id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2502.09591
Censor Dependent Variational Inference
cs.LG stat.ML
This paper provides a comprehensive analysis of variational inference in latent variable models for survival analysis, emphasizing the distinctive challenges associated with applying variational methods to survival data. We identify a critical weakness in the existing methodology, demonstrating how a poorly designed variational distribution may hinder the objective of survival analysis tasks--modeling time-to-event distributions. We prove that the optimal variational distribution, which perfectly bounds the log-likelihood, may depend on the censoring mechanism. To address this issue, we propose censor-dependent variational inference (CDVI), tailored for latent variable models in survival analysis. More practically, we introduce CD-CVAE, a V-structure Variational Autoencoder (VAE) designed for the scalable implementation of CDVI. Further discussion extends some existing theories and training techniques to survival analysis. Extensive experiments validate our analysis and demonstrate significant improvements in the estimation of individual survival distributions.
2502.09592
A Data-Driven Method for Microgrid System Identification: Physically Consistent Sparse Identification of Nonlinear Dynamics
eess.SY cs.SY
Microgrids (MGs) play a crucial role in utilizing distributed energy resources (DERs) like solar and wind power, enhancing the sustainability and flexibility of modern power systems. However, the inherent variability in MG topology, power flow, and DER operating modes poses significant challenges to the accurate system identification of MGs, which is crucial for designing robust control strategies and ensuring MG stability. This paper proposes a Physically Consistent Sparse Identification of Nonlinear Dynamics (PC-SINDy) method for accurate MG system identification. By leveraging an analytically derived library of candidate functions, PC-SINDy extracts accurate dynamic models using only phasor measurement unit (PMU) data. Simulations on a 4-bus system demonstrate that PC-SINDy can reliably and accurately predict frequency trajectories under large disturbances, including scenarios not encountered during the identification/training phase, even when using noisy, low-sampled PMU data.
2502.09596
KIMAs: A Configurable Knowledge Integrated Multi-Agent System
cs.AI cs.MA
Knowledge-intensive conversations supported by large language models (LLMs) have become one of the most popular and helpful applications that can assist people in different aspects. Many current knowledge-intensive applications are centered on retrieval-augmented generation (RAG) techniques. While many open-source RAG frameworks facilitate the development of RAG-based applications, they often fall short in handling practical scenarios complicated by heterogeneous data in topics and formats, conversational context management, and the requirement of low-latency response times. This technical report presents a configurable knowledge integrated multi-agent system, KIMAs, to address these challenges. KIMAs features a flexible and configurable system for integrating diverse knowledge sources with 1) context management and query rewrite mechanisms to improve retrieval accuracy and multi-turn conversational coherency, 2) efficient knowledge routing and retrieval, 3) simple but effective filter and reference generation mechanisms, and 4) optimized parallelizable multi-agent pipeline execution. Our work provides a scalable framework for advancing the deployment of LLMs in real-world settings. To show how KIMAs can help developers build knowledge-intensive applications with different scales and emphases, we demonstrate how we configure the system to three applications already running in practice with reliable performance.
2502.09597
Do LLMs Recognize Your Preferences? Evaluating Personalized Preference Following in LLMs
cs.LG cs.CL
Large Language Models (LLMs) are increasingly used as chatbots, yet their ability to personalize responses to user preferences remains limited. We introduce PrefEval, a benchmark for evaluating LLMs' ability to infer, memorize and adhere to user preferences in a long-context conversational setting. PrefEval comprises 3,000 manually curated user preference and query pairs spanning 20 topics. PrefEval contains user personalization or preference information in both explicit and implicit forms, and evaluates LLM performance using a generation and a classification task. With PrefEval, we evaluated the aforementioned preference following capabilities of 10 open-source and proprietary LLMs in multi-session conversations with varying context lengths up to 100k tokens. We benchmark with various prompting, iterative feedback, and retrieval-augmented generation methods. Our benchmarking effort reveals that state-of-the-art LLMs face significant challenges in proactively following users' preferences during conversations. In particular, in zero-shot settings, preference following accuracy falls below 10% at merely 10 turns (~3k tokens) across most evaluated models. Even with advanced prompting and retrieval methods, preference following still deteriorates in long-context conversations. Furthermore, we show that fine-tuning on PrefEval significantly improves performance. We believe PrefEval serves as a valuable resource for measuring, understanding, and enhancing LLMs' preference following abilities, paving the way for personalized conversational agents. Our code and dataset are available at https://prefeval.github.io/.
2502.09598
GAIA: A Global, Multi-modal, Multi-scale Vision-Language Dataset for Remote Sensing Image Analysis
cs.CV
The continuous operation of Earth-orbiting satellites generates vast and ever-growing archives of Remote Sensing (RS) images. Natural language presents an intuitive interface for accessing, querying, and interpreting the data from such archives. However, existing Vision-Language Models (VLMs) are predominantly trained on web-scraped, noisy image-text data, exhibiting limited exposure to the specialized domain of RS. This deficiency results in poor performance on RS-specific tasks, as commonly used datasets often lack detailed, scientifically accurate textual descriptions and instead emphasize solely on attributes like date and location. To bridge this critical gap, we introduce GAIA, a novel dataset designed for multi-scale, multi-sensor, and multi-modal RS image analysis. GAIA comprises of 205,150 meticulously curated RS image-text pairs, representing a diverse range of RS modalities associated to different spatial resolutions. Unlike existing vision-language datasets in RS, GAIA specifically focuses on capturing a diverse range of RS applications, providing unique information about environmental changes, natural disasters, and various other dynamic phenomena. The dataset provides a spatially and temporally balanced distribution, spanning across the globe, covering the last 25 years with a balanced temporal distribution of observations. GAIA's construction involved a two-stage process: (1) targeted web-scraping of images and accompanying text from reputable RS-related sources, and (2) generation of five high-quality, scientifically grounded synthetic captions for each image using carefully crafted prompts that leverage the advanced vision-language capabilities of GPT-4o. Our extensive experiments, including fine-tuning of CLIP and BLIP2 models, demonstrate that GAIA significantly improves performance on RS image classification, cross-modal retrieval and image captioning tasks.
2502.09601
CoT-Valve: Length-Compressible Chain-of-Thought Tuning
cs.AI cs.CL
Chain-of-Thought significantly enhances a model's reasoning capability, but it also comes with a considerable increase in inference costs due to long chains. With the observation that the reasoning path can be easily compressed under easy tasks but struggle on hard tasks, we explore the feasibility of elastically controlling the length of reasoning paths with only one model, thereby reducing the inference overhead of reasoning models dynamically based on task difficulty. We introduce a new tuning and inference strategy named CoT-Valve, designed to allow models to generate reasoning chains of varying lengths. To achieve this, we propose to identify a direction in the parameter space that, when manipulated, can effectively control the length of generated CoT. Moreover, we show that this property is valuable for compressing the reasoning chain. We construct datasets with chains from long to short for the same questions and explore two enhanced strategies for CoT-Valve: (1) a precise length-compressible CoT tuning method, and (2) a progressive chain length compression approach. Our experiments show that CoT-Valve successfully enables controllability and compressibility of the chain and shows better performance than the prompt-based control. We applied this method to QwQ-32B-Preview, reducing reasoning chains on GSM8K from 741 to 225 tokens with a minor performance drop (95.07% to 94.92%) and on AIME from 6827 to 4629 tokens, with only one additional incorrect answer.
2502.09604
SelfCite: Self-Supervised Alignment for Context Attribution in Large Language Models
cs.CL cs.AI cs.LG
We introduce SelfCite, a novel self-supervised approach that aligns LLMs to generate high-quality, fine-grained, sentence-level citations for the statements in their generated responses. Instead of only relying on costly and labor-intensive annotations, SelfCite leverages a reward signal provided by the LLM itself through context ablation: If a citation is necessary, removing the cited text from the context should prevent the same response; if sufficient, retaining the cited text alone should preserve the same response. This reward can guide the inference-time best-of-N sampling strategy to improve citation quality significantly, as well as be used in preference optimization to directly fine-tune the models for generating better citations. The effectiveness of SelfCite is demonstrated by increasing citation F1 up to 5.3 points on the LongBench-Cite benchmark across five long-form question answering tasks.
2502.09606
Human-LLM Coevolution: Evidence from Academic Writing
cs.CL cs.AI cs.CY cs.DL cs.LG
With a statistical analysis of arXiv paper abstracts, we report a marked drop in the frequency of several words previously identified as overused by ChatGPT, such as "delve", starting soon after they were pointed out in early 2024. The frequency of certain other words favored by ChatGPT, such as "significant", has instead kept increasing. These phenomena suggest that some authors of academic papers have adapted their use of large language models (LLMs), for example, by selecting outputs or applying modifications to the LLM-generated content. Such coevolution and cooperation of humans and LLMs thus introduce additional challenges to the detection of machine-generated text in real-world scenarios. Estimating the impact of LLMs on academic writing by examining word frequency remains feasible, and more attention should be paid to words that were already frequently employed, including those that have decreased in frequency due to LLMs' disfavor.
2502.09608
Instance Segmentation of Scene Sketches Using Natural Image Priors
cs.CV cs.GR
Sketch segmentation involves grouping pixels within a sketch that belong to the same object or instance. It serves as a valuable tool for sketch editing tasks, such as moving, scaling, or removing specific components. While image segmentation models have demonstrated remarkable capabilities in recent years, sketches present unique challenges for these models due to their sparse nature and wide variation in styles. We introduce SketchSeg, a method for instance segmentation of raster scene sketches. Our approach adapts state-of-the-art image segmentation and object detection models to the sketch domain by employing class-agnostic fine-tuning and refining segmentation masks using depth cues. Furthermore, our method organizes sketches into sorted layers, where occluded instances are inpainted, enabling advanced sketch editing applications. As existing datasets in this domain lack variation in sketch styles, we construct a synthetic scene sketch segmentation dataset featuring sketches with diverse brush strokes and varying levels of detail. We use this dataset to demonstrate the robustness of our approach and will release it to promote further research in the field. Project webpage: https://sketchseg.github.io/sketch-seg/
2502.09609
Score-of-Mixture Training: Training One-Step Generative Models Made Simple via Score Estimation of Mixture Distributions
cs.LG cs.AI stat.ML
We propose Score-of-Mixture Training (SMT), a novel framework for training one-step generative models by minimizing a class of divergences called the $\alpha$-skew Jensen-Shannon divergence. At its core, SMT estimates the score of mixture distributions between real and fake samples across multiple noise levels. Similar to consistency models, our approach supports both training from scratch (SMT) and distillation using a pretrained diffusion model, which we call Score-of-Mixture Distillation (SMD). It is simple to implement, requires minimal hyperparameter tuning, and ensures stable training. Experiments on CIFAR-10 and ImageNet 64x64 show that SMT/SMD are competitive with and can even outperform existing methods.
2502.09611
Designing a Conditional Prior Distribution for Flow-Based Generative Models
cs.LG cs.CV
Flow-based generative models have recently shown impressive performance for conditional generation tasks, such as text-to-image generation. However, current methods transform a general unimodal noise distribution to a specific mode of the target data distribution. As such, every point in the initial source distribution can be mapped to every point in the target distribution, resulting in long average paths. To this end, in this work, we tap into a non-utilized property of conditional flow-based models: the ability to design a non-trivial prior distribution. Given an input condition, such as a text prompt, we first map it to a point lying in data space, representing an ``average" data point with the minimal average distance to all data points of the same conditional mode (e.g., class). We then utilize the flow matching formulation to map samples from a parametric distribution centered around this point to the conditional target distribution. Experimentally, our method significantly improves training times and generation efficiency (FID, KID and CLIP alignment scores) compared to baselines, producing high quality samples using fewer sampling steps.
2502.09613
Latent Radiance Fields with 3D-aware 2D Representations
cs.CV
Latent 3D reconstruction has shown great promise in empowering 3D semantic understanding and 3D generation by distilling 2D features into the 3D space. However, existing approaches struggle with the domain gap between 2D feature space and 3D representations, resulting in degraded rendering performance. To address this challenge, we propose a novel framework that integrates 3D awareness into the 2D latent space. The framework consists of three stages: (1) a correspondence-aware autoencoding method that enhances the 3D consistency of 2D latent representations, (2) a latent radiance field (LRF) that lifts these 3D-aware 2D representations into 3D space, and (3) a VAE-Radiance Field (VAE-RF) alignment strategy that improves image decoding from the rendered 2D representations. Extensive experiments demonstrate that our method outperforms the state-of-the-art latent 3D reconstruction approaches in terms of synthesis performance and cross-dataset generalizability across diverse indoor and outdoor scenes. To our knowledge, this is the first work showing the radiance field representations constructed from 2D latent representations can yield photorealistic 3D reconstruction performance.
2502.09614
DexTrack: Towards Generalizable Neural Tracking Control for Dexterous Manipulation from Human References
cs.RO cs.AI cs.CV cs.LG
We address the challenge of developing a generalizable neural tracking controller for dexterous manipulation from human references. This controller aims to manage a dexterous robot hand to manipulate diverse objects for various purposes defined by kinematic human-object interactions. Developing such a controller is complicated by the intricate contact dynamics of dexterous manipulation and the need for adaptivity, generalizability, and robustness. Current reinforcement learning and trajectory optimization methods often fall short due to their dependence on task-specific rewards or precise system models. We introduce an approach that curates large-scale successful robot tracking demonstrations, comprising pairs of human references and robot actions, to train a neural controller. Utilizing a data flywheel, we iteratively enhance the controller's performance, as well as the number and quality of successful tracking demonstrations. We exploit available tracking demonstrations and carefully integrate reinforcement learning and imitation learning to boost the controller's performance in dynamic environments. At the same time, to obtain high-quality tracking demonstrations, we individually optimize per-trajectory tracking by leveraging the learned tracking controller in a homotopy optimization method. The homotopy optimization, mimicking chain-of-thought, aids in solving challenging trajectory tracking problems to increase demonstration diversity. We showcase our success by training a generalizable neural controller and evaluating it in both simulation and real world. Our method achieves over a 10% improvement in success rates compared to leading baselines. The project website with animated results is available at https://meowuu7.github.io/DexTrack/.
2502.09615
RigAnything: Template-Free Autoregressive Rigging for Diverse 3D Assets
cs.CV
We present RigAnything, a novel autoregressive transformer-based model, which makes 3D assets rig-ready by probabilistically generating joints, skeleton topologies, and assigning skinning weights in a template-free manner. Unlike most existing auto-rigging methods, which rely on predefined skeleton template and are limited to specific categories like humanoid, RigAnything approaches the rigging problem in an autoregressive manner, iteratively predicting the next joint based on the global input shape and the previous prediction. While autoregressive models are typically used to generate sequential data, RigAnything extends their application to effectively learn and represent skeletons, which are inherently tree structures. To achieve this, we organize the joints in a breadth-first search (BFS) order, enabling the skeleton to be defined as a sequence of 3D locations and the parent index. Furthermore, our model improves the accuracy of position prediction by leveraging diffusion modeling, ensuring precise and consistent placement of joints within the hierarchy. This formulation allows the autoregressive model to efficiently capture both spatial and hierarchical relationships within the skeleton. Trained end-to-end on both RigNet and Objaverse datasets, RigAnything demonstrates state-of-the-art performance across diverse object types, including humanoids, quadrupeds, marine creatures, insects, and many more, surpassing prior methods in quality, robustness, generalizability, and efficiency. Please check our website for more details: https://www.liuisabella.com/RigAnything.
2502.09616
Variational Rectified Flow Matching
cs.LG cs.CV
We study Variational Rectified Flow Matching, a framework that enhances classic rectified flow matching by modeling multi-modal velocity vector-fields. At inference time, classic rectified flow matching 'moves' samples from a source distribution to the target distribution by solving an ordinary differential equation via integration along a velocity vector-field. At training time, the velocity vector-field is learnt by linearly interpolating between coupled samples one drawn from the source and one drawn from the target distribution randomly. This leads to ''ground-truth'' velocity vector-fields that point in different directions at the same location, i.e., the velocity vector-fields are multi-modal/ambiguous. However, since training uses a standard mean-squared-error loss, the learnt velocity vector-field averages ''ground-truth'' directions and isn't multi-modal. In contrast, variational rectified flow matching learns and samples from multi-modal flow directions. We show on synthetic data, MNIST, CIFAR-10, and ImageNet that variational rectified flow matching leads to compelling results.
2502.09617
LIFe-GoM: Generalizable Human Rendering with Learned Iterative Feedback Over Multi-Resolution Gaussians-on-Mesh
cs.CV
Generalizable rendering of an animatable human avatar from sparse inputs relies on data priors and inductive biases extracted from training on large data to avoid scene-specific optimization and to enable fast reconstruction. This raises two main challenges: First, unlike iterative gradient-based adjustment in scene-specific optimization, generalizable methods must reconstruct the human shape representation in a single pass at inference time. Second, rendering is preferably computationally efficient yet of high resolution. To address both challenges we augment the recently proposed dual shape representation, which combines the benefits of a mesh and Gaussian points, in two ways. To improve reconstruction, we propose an iterative feedback update framework, which successively improves the canonical human shape representation during reconstruction. To achieve computationally efficient yet high-resolution rendering, we study a coupled-multi-resolution Gaussians-on-Mesh representation. We evaluate the proposed approach on the challenging THuman2.0, XHuman and AIST++ data. Our approach reconstructs an animatable representation from sparse inputs in less than 1s, renders views with 95.1FPS at $1024 \times 1024$, and achieves PSNR/LPIPS*/FID of 24.65/110.82/51.27 on THuman2.0, outperforming the state-of-the-art in rendering quality.
2502.09619
Can this Model Also Recognize Dogs? Zero-Shot Model Search from Weights
cs.LG cs.CV
With the increasing numbers of publicly available models, there are probably pretrained, online models for most tasks users require. However, current model search methods are rudimentary, essentially a text-based search in the documentation, thus users cannot find the relevant models. This paper presents ProbeLog, a method for retrieving classification models that can recognize a target concept, such as "Dog", without access to model metadata or training data. Differently from previous probing methods, ProbeLog computes a descriptor for each output dimension (logit) of each model, by observing its responses on a fixed set of inputs (probes). Our method supports both logit-based retrieval ("find more logits like this") and zero-shot, text-based retrieval ("find all logits corresponding to dogs"). As probing-based representations require multiple costly feedforward passes through the model, we develop a method, based on collaborative filtering, that reduces the cost of encoding repositories by 3x. We demonstrate that ProbeLog achieves high retrieval accuracy, both in real-world and fine-grained search tasks and is scalable to full-size repositories.
2502.09620
Exploring the Potential of Encoder-free Architectures in 3D LMMs
cs.CV cs.AI cs.CL
Encoder-free architectures have been preliminarily explored in the 2D visual domain, yet it remains an open question whether they can be effectively applied to 3D understanding scenarios. In this paper, we present the first comprehensive investigation into the potential of encoder-free architectures to overcome the challenges of encoder-based 3D Large Multimodal Models (LMMs). These challenges include the failure to adapt to varying point cloud resolutions and the point features from the encoder not meeting the semantic needs of Large Language Models (LLMs). We identify key aspects for 3D LMMs to remove the encoder and enable the LLM to assume the role of the 3D encoder: 1) We propose the LLM-embedded Semantic Encoding strategy in the pre-training stage, exploring the effects of various point cloud self-supervised losses. And we present the Hybrid Semantic Loss to extract high-level semantics. 2) We introduce the Hierarchical Geometry Aggregation strategy in the instruction tuning stage. This incorporates inductive bias into the LLM early layers to focus on the local details of the point clouds. To the end, we present the first Encoder-free 3D LMM, ENEL. Our 7B model rivals the current state-of-the-art model, ShapeLLM-13B, achieving 55.0%, 50.92%, and 42.7% on the classification, captioning, and VQA tasks, respectively. Our results demonstrate that the encoder-free architecture is highly promising for replacing encoder-based architectures in the field of 3D understanding. The code is released at https://github.com/Ivan-Tang-3D/ENEL
2502.09621
MME-CoT: Benchmarking Chain-of-Thought in Large Multimodal Models for Reasoning Quality, Robustness, and Efficiency
cs.CV cs.AI cs.CL
Answering questions with Chain-of-Thought (CoT) has significantly enhanced the reasoning capabilities of Large Language Models (LLMs), yet its impact on Large Multimodal Models (LMMs) still lacks a systematic assessment and in-depth investigation. In this paper, we introduce MME-CoT, a specialized benchmark evaluating the CoT reasoning performance of LMMs, spanning six domains: math, science, OCR, logic, space-time, and general scenes. As the first comprehensive study in this area, we propose a thorough evaluation suite incorporating three novel metrics that assess the reasoning quality, robustness, and efficiency at a fine-grained level. Leveraging curated high-quality data and a unique evaluation strategy, we conduct an in-depth analysis of state-of-the-art LMMs, uncovering several key insights: 1) Models with reflection mechanism demonstrate a superior CoT quality, with Kimi k1.5 outperforming GPT-4o and demonstrating the highest quality results; 2) CoT prompting often degrades LMM performance on perception-heavy tasks, suggesting a potentially harmful overthinking behavior; and 3) Although the CoT quality is high, LMMs with reflection exhibit significant inefficiency in both normal response and self-correction phases. We hope MME-CoT serves as a foundation for advancing multimodal reasoning in LMMs. Project Page: https://mmecot.github.io/
2502.09622
Theoretical Benefit and Limitation of Diffusion Language Model
cs.LG cs.AI cs.CL stat.ML
Diffusion language models have emerged as a promising approach for text generation. One would naturally expect this method to be an efficient replacement for autoregressive models since multiple tokens can be sampled in parallel during each diffusion step. However, its efficiency-accuracy trade-off is not yet well understood. In this paper, we present a rigorous theoretical analysis of a widely used type of diffusion language model, the Masked Diffusion Model (MDM), and find that its effectiveness heavily depends on the target evaluation metric. Under mild conditions, we prove that when using perplexity as the metric, MDMs can achieve near-optimal perplexity in sampling steps regardless of sequence length, demonstrating that efficiency can be achieved without sacrificing performance. However, when using the sequence error rate--which is important for understanding the "correctness" of a sequence, such as a reasoning chain--we show that the required sampling steps must scale linearly with sequence length to obtain "correct" sequences, thereby eliminating MDM's efficiency advantage over autoregressive models. Our analysis establishes the first theoretical foundation for understanding the benefits and limitations of MDMs. All theoretical findings are supported by empirical studies.
2502.09623
Embed Any NeRF: Graph Meta-Networks for Neural Tasks on Arbitrary NeRF Architectures
cs.CV
Neural Radiance Fields (NeRFs) have emerged as a groundbreaking paradigm for representing 3D objects and scenes by encoding shape and appearance information into the weights of a neural network. Recent works have shown how such weights can be used as input to frameworks processing them to solve deep learning tasks. Yet, these frameworks can only process NeRFs with a specific, predefined architecture. In this paper, we present the first framework that can ingest NeRFs with multiple architectures and perform inference on architectures unseen at training time. We achieve this goal by training a Graph Meta-Network in a representation learning framework. Moreover, we show how a contrastive objective is conducive to obtaining an architecture-agnostic latent space. In experiments on both MLP-based and tri-planar NeRFs, our approach demonstrates robust performance in classification and retrieval tasks that either matches or exceeds that of existing frameworks constrained to single architectures, thus providing the first architecture-agnostic method to perform tasks on NeRFs by processing their weights.
2502.09624
Efficient and Trustworthy Block Propagation for Blockchain-enabled Mobile Embodied AI Networks: A Graph Resfusion Approach
cs.AI cs.CR
By synergistically integrating mobile networks and embodied artificial intelligence (AI), Mobile Embodied AI Networks (MEANETs) represent an advanced paradigm that facilitates autonomous, context-aware, and interactive behaviors within dynamic environments. Nevertheless, the rapid development of MEANETs is accompanied by challenges in trustworthiness and operational efficiency. Fortunately, blockchain technology, with its decentralized and immutable characteristics, offers promising solutions for MEANETs. However, existing block propagation mechanisms suffer from challenges such as low propagation efficiency and weak security for block propagation, which results in delayed transmission of vehicular messages or vulnerability to malicious tampering, potentially causing severe traffic accidents in blockchain-enabled MEANETs. Moreover, current block propagation strategies cannot effectively adapt to real-time changes of dynamic topology in MEANETs. Therefore, in this paper, we propose a graph Resfusion model-based trustworthy block propagation optimization framework for consortium blockchain-enabled MEANETs. Specifically, we propose an innovative trust calculation mechanism based on the trust cloud model, which comprehensively accounts for randomness and fuzziness in the miner trust evaluation. Furthermore, by leveraging the strengths of graph neural networks and diffusion models, we develop a graph Resfusion model to effectively and adaptively generate the optimal block propagation trajectory. Simulation results demonstrate that the proposed model outperforms other routing mechanisms in terms of block propagation efficiency and trustworthiness. Additionally, the results highlight its strong adaptability to dynamic environments, making it particularly suitable for rapidly changing MEANETs.
2502.09625
Transformer Based Time-Series Forecasting for Stock
q-fin.CP cs.LG
To the naked eye, stock prices are considered chaotic, dynamic, and unpredictable. Indeed, it is one of the most difficult forecasting tasks that hundreds of millions of retail traders and professional traders around the world try to do every second even before the market opens. With recent advances in the development of machine learning and the amount of data the market generated over years, applying machine learning techniques such as deep learning neural networks is unavoidable. In this work, we modeled the task as a multivariate forecasting problem, instead of a naive autoregression problem. The multivariate analysis is done using the attention mechanism via applying a mutated version of the Transformer, "Stockformer", which we created.
2502.09626
On the Bias, Fairness, and Bias Mitigation for a Wearable-based Freezing of Gait Detection in Parkinson's Disease
eess.SP cs.LG
Freezing of gait (FOG) is a debilitating feature of Parkinson's disease (PD), which is a cause of injurious falls among PD patients. Recent advances in wearable-based human activity recognition (HAR) technology have enabled the detection of FOG subtypes across benchmark datasets. Since FOG manifestation is heterogeneous, developing models that quantify FOG consistently across patients with varying demographics, FOG types, and PD conditions is important. Bias and fairness in FOG models remain understudied in HAR, with research focused mainly on FOG detection using single benchmark datasets. We evaluated the bias and fairness of HAR models for wearable-based FOG detection across demographics and PD conditions using multiple datasets and the effectiveness of transfer learning as a potential bias mitigation approach. Our evaluation using demographic parity ratio (DPR) and equalized odds ratio (EOR) showed model bias (DPR & EOR < 0.8) for all stratified demographic variables, including age, sex, and disease duration. Our experiments demonstrated that transfer learning from multi-site datasets and generic human activity representations significantly improved fairness (average change in DPR +0.027, +0.039, respectively) and performance (average change in F1-score +0.026, +0.018, respectively) across attributes, supporting the hypothesis that generic human activity representations learn fairer representations applicable to health analytics.
2502.09635
CORRECT: Context- and Reference-Augmented Reasoning and Prompting for Fact-Checking
cs.CL cs.AI
Fact-checking the truthfulness of claims usually requires reasoning over multiple evidence sentences. Oftentimes, evidence sentences may not be always self-contained, and may require additional contexts and references from elsewhere to understand coreferential expressions, acronyms, and the scope of a reported finding. For example, evidence sentences from an academic paper may need contextual sentences in the paper and descriptions in its cited papers to determine the scope of a research discovery. However, most fact-checking models mainly focus on the reasoning within evidence sentences, and ignore the auxiliary contexts and references. To address this problem, we propose a novel method, Context- and Reference-augmented Reasoning and Prompting. For evidence reasoning, we construct a three-layer evidence graph with evidence, context, and reference layers. We design intra- and cross-layer reasoning to integrate three graph layers into a unified evidence embedding. For verdict prediction, we design evidence-conditioned prompt encoder, which produces unique prompt embeddings for each claim. These evidence-conditioned prompt embeddings and claims are unified for fact-checking. Experiments verify the strength of our model.
2502.09636
Reading between the Lines: Can LLMs Identify Cross-Cultural Communication Gaps?
cs.CL cs.AI
In a rapidly globalizing and digital world, content such as book and product reviews created by people from diverse cultures are read and consumed by others from different corners of the world. In this paper, we investigate the extent and patterns of gaps in understandability of book reviews due to the presence of culturally-specific items and elements that might be alien to users from another culture. Our user-study on 57 book reviews from Goodreads reveal that 83\% of the reviews had at least one culture-specific difficult-to-understand element. We also evaluate the efficacy of GPT-4o in identifying such items, given the cultural background of the reader; the results are mixed, implying a significant scope for improvement. Our datasets are available here: https://github.com/sougata-ub/reading_between_lines
2502.09637
Meta-Cultural Competence: Climbing the Right Hill of Cultural Awareness
cs.CY cs.AI cs.CL
Numerous recent studies have shown that Large Language Models (LLMs) are biased towards a Western and Anglo-centric worldview, which compromises their usefulness in non-Western cultural settings. However, "culture" is a complex, multifaceted topic, and its awareness, representation, and modeling in LLMs and LLM-based applications can be defined and measured in numerous ways. In this position paper, we ask what does it mean for an LLM to possess "cultural awareness", and through a thought experiment, which is an extension of the Octopus test proposed by Bender and Koller (2020), we argue that it is not cultural awareness or knowledge, rather meta-cultural competence, which is required of an LLM and LLM-based AI system that will make it useful across various, including completely unseen, cultures. We lay out the principles of meta-cultural competence AI systems, and discuss ways to measure and model those.
2502.09638
Jailbreaking to Jailbreak
cs.CL cs.AI
Refusal training on Large Language Models (LLMs) prevents harmful outputs, yet this defense remains vulnerable to both automated and human-crafted jailbreaks. We present a novel LLM-as-red-teamer approach in which a human jailbreaks a refusal-trained LLM to make it willing to jailbreak itself or other LLMs. We refer to the jailbroken LLMs as $J_2$ attackers, which can systematically evaluate target models using various red teaming strategies and improve its performance via in-context learning from the previous failures. Our experiments demonstrate that Sonnet 3.5 and Gemini 1.5 pro outperform other LLMs as $J_2$, achieving 93.0% and 91.0% attack success rates (ASRs) respectively against GPT-4o (and similar results across other capable LLMs) on Harmbench. Our work not only introduces a scalable approach to strategic red teaming, drawing inspiration from human red teamers, but also highlights jailbreaking-to-jailbreak as an overlooked failure mode of the safeguard. Specifically, an LLM can bypass its own safeguards by employing a jailbroken version of itself that is willing to assist in further jailbreaking. To prevent any direct misuse with $J_2$, while advancing research in AI safety, we publicly share our methodology while keeping specific prompting details private.
2502.09640
Online Social Support Detection in Spanish Social Media Texts
cs.CL cs.AI
The advent of social media has transformed communication, enabling individuals to share their experiences, seek support, and participate in diverse discussions. While extensive research has focused on identifying harmful content like hate speech, the recognition and promotion of positive and supportive interactions remain largely unexplored. This study proposes an innovative approach to detecting online social support in Spanish-language social media texts. We introduce the first annotated dataset specifically created for this task, comprising 3,189 YouTube comments classified as supportive or non-supportive. To address data imbalance, we employed GPT-4o to generate paraphrased comments and create a balanced dataset. We then evaluated social support classification using traditional machine learning models, deep learning architectures, and transformer-based models, including GPT-4o, but only on the unbalanced dataset. Subsequently, we utilized a transformer model to compare the performance between the balanced and unbalanced datasets. Our findings indicate that the balanced dataset yielded improved results for Task 2 (Individual and Group) and Task 3 (Nation, Other, LGBTQ, Black Community, Women, Religion), whereas GPT-4o performed best for Task 1 (Social Support and Non-Support). This study highlights the significance of fostering a supportive online environment and lays the groundwork for future research in automated social support detection.
2502.09642
Krutrim LLM: Multilingual Foundational Model for over a Billion People
cs.CL cs.AI
India is a diverse society with unique challenges in developing AI systems, including linguistic diversity, oral traditions, data accessibility, and scalability. Existing foundation models are primarily trained on English, limiting their effectiveness for India's population. Indic languages comprise only 1 percent of Common Crawl corpora despite India representing 18 percent of the global population, leading to linguistic biases. Thousands of regional languages, dialects, and code mixing create additional representation challenges due to sparse training data. We introduce Krutrim LLM, a 2 trillion token multilingual model designed for India's linguistic landscape. It incorporates the largest known Indic dataset, mitigating data scarcity and ensuring balanced performance across dialects. Krutrim outperforms or matches state-of-the-art models on Indic benchmarks while maintaining competitive English performance. Despite being significantly smaller in training flops, Krutrim LLM matches or exceeds models like LLAMA-2 on 10 out of 16 tasks, with an average score of 0.57 versus 0.55. This evidences Krutrim's flexible multilingual fluency across diverse linguistic contexts. Krutrim is integrated with real-time search to improve factual accuracy in conversational AI applications. This enhances accessibility for over 1 billion users worldwide. Through intentional design choices addressing data imbalances, Krutrim LLM signifies meaningful progress in building ethical, globally representative AI models.
2502.09644
From Argumentation to Deliberation: Perspectivized Stance Vectors for Fine-grained (Dis)agreement Analysis
cs.CL cs.AI cs.CY
Debating over conflicting issues is a necessary first step towards resolving conflicts. However, intrinsic perspectives of an arguer are difficult to overcome by persuasive argumentation skills. Proceeding from a debate to a deliberative process, where we can identify actionable options for resolving a conflict requires a deeper analysis of arguments and the perspectives they are grounded in - as it is only from there that one can derive mutually agreeable resolution steps. In this work we develop a framework for a deliberative analysis of arguments in a computational argumentation setup. We conduct a fine-grained analysis of perspectivized stances expressed in the arguments of different arguers or stakeholders on a given issue, aiming not only to identify their opposing views, but also shared perspectives arising from their attitudes, values or needs. We formalize this analysis in Perspectivized Stance Vectors that characterize the individual perspectivized stances of all arguers on a given issue. We construct these vectors by determining issue- and argument-specific concepts, and predict an arguer's stance relative to each of them. The vectors allow us to measure a modulated (dis)agreement between arguers, structured by perspectives, which allows us to identify actionable points for conflict resolution, as a first step towards deliberation.
2502.09645
From No to Know: Taxonomy, Challenges, and Opportunities for Negation Understanding in Multimodal Foundation Models
cs.CL cs.AI
Negation, a linguistic construct conveying absence, denial, or contradiction, poses significant challenges for multilingual multimodal foundation models. These models excel in tasks like machine translation, text-guided generation, image captioning, audio interactions, and video processing but often struggle to accurately interpret negation across diverse languages and cultural contexts. In this perspective paper, we propose a comprehensive taxonomy of negation constructs, illustrating how structural, semantic, and cultural factors influence multimodal foundation models. We present open research questions and highlight key challenges, emphasizing the importance of addressing these issues to achieve robust negation handling. Finally, we advocate for specialized benchmarks, language-specific tokenization, fine-grained attention mechanisms, and advanced multimodal architectures. These strategies can foster more adaptable and semantically precise multimodal foundation models, better equipped to navigate and accurately interpret the complexities of negation in multilingual, multimodal environments.
2502.09646
Language Shift or Maintenance? An Intergenerational Study of the Tibetan Community in Saudi Arabia
cs.CL cs.CY
The present study provides the first-ever report on the language shift from Tibetan to Arabic among descendants of Tibetan families who migrated from the Tibet region to Saudi Arabia around 70 years ago. The aim of this study was to determine whether three age groups had adopted different practices in terms of maintaining Tibetan or shifting to Hijazi Arabic. To this end, 96 male and female members of the Tibetan community responded to a questionnaire in which they were asked about their code choice in different domains (home, neighbourhood, friends and relatives, expressing emotion, and performing religious rituals). The data revealed significant intergenerational differences between members of the community in terms of the extent of the shift to Arabic, with Tibetan rarely used by younger members and older members making only slightly more use of it. The difference between the three age groups was significant, at a p-value of .001.
2502.09647
Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification
cs.CL cs.LG
The ability to process long contexts is crucial for many natural language processing tasks, yet it remains a significant challenge. While substantial progress has been made in enhancing the efficiency of attention mechanisms, there is still a gap in understanding how attention heads function in long-context settings. In this paper, we observe that while certain heads consistently attend to local information only, others swing between attending to local and long-context information depending on the query. This raises the question: can we identify which heads require long-context information to predict the next token accurately? We demonstrate that it's possible to predict which heads are crucial for long-context processing using only local keys. The core idea here is to exploit a simple model for the long-context scores via second moment approximations. These findings unveil simple properties of attention in the context of long sequences, and open the door to potentially significant gains in efficiency.
2502.09648
UKTA: Unified Korean Text Analyzer
cs.CL cs.AI
Evaluating writing quality is complex and time-consuming often delaying feedback to learners. While automated writing evaluation tools are effective for English, Korean automated writing evaluation tools face challenges due to their inability to address multi-view analysis, error propagation, and evaluation explainability. To overcome these challenges, we introduce UKTA (Unified Korean Text Analyzer), a comprehensive Korea text analysis and writing evaluation system. UKTA provides accurate low-level morpheme analysis, key lexical features for mid-level explainability, and transparent high-level rubric-based writing scores. Our approach enhances accuracy and quadratic weighted kappa over existing baseline, positioning UKTA as a leading multi-perspective tool for Korean text analysis and writing evaluation.
2502.09649
Imit Diff: Semantics Guided Diffusion Transformer with Dual Resolution Fusion for Imitation Learning
cs.AI cs.CV cs.LG cs.RO
Visuomotor imitation learning enables embodied agents to effectively acquire manipulation skills from video demonstrations and robot proprioception. However, as scene complexity and visual distractions increase, existing methods that perform well in simple scenes tend to degrade in performance. To address this challenge, we introduce Imit Diff, a semanstic guided diffusion transformer with dual resolution fusion for imitation learning. Our approach leverages prior knowledge from vision language foundation models to translate high-level semantic instruction into pixel-level visual localization. This information is explicitly integrated into a multi-scale visual enhancement framework, constructed with a dual resolution encoder. Additionally, we introduce an implementation of Consistency Policy within the diffusion transformer architecture to improve both real-time performance and motion smoothness in embodied agent control.We evaluate Imit Diff on several challenging real-world tasks. Due to its task-oriented visual localization and fine-grained scene perception, it significantly outperforms state-of-the-art methods, especially in complex scenes with visual distractions, including zero-shot experiments focused on visual distraction and category generalization. The code will be made publicly available.
2502.09650
Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples
cs.CL cs.AI cs.LG
The alignment of large language models (LLMs) often assumes that using more clean data yields better outcomes, overlooking the match between model capacity and example difficulty. Challenging this, we propose a new principle: Preference data vary in difficulty, and overly difficult examples hinder alignment, by exceeding the model's capacity. Through systematic experimentation, we validate this principle with three key findings: (1) preference examples vary in difficulty, as evidenced by consistent learning orders across alignment runs; (2) overly difficult examples significantly degrade performance across four LLMs and two datasets; and (3) the capacity of a model dictates its threshold for handling difficult examples, underscoring a critical relationship between data selection and model capacity. Building on this principle, we introduce Selective DPO, which filters out overly difficult examples. This simple adjustment improves alignment performance by 9-16% in win rates on the AlpacaEval 2 benchmark compared to the DPO baseline, suppressing a series of DPO variants with different algorithmic adjustments. Together, these results illuminate the importance of aligning data difficulty with model capacity, offering a transformative perspective for improving alignment strategies in LLMs. Code is available at https://github.com/glorgao/SelectiveDPO.
2502.09651
AI-VERDE: A Gateway for Egalitarian Access to Large Language Model-Based Resources For Educational Institutions
cs.CL cs.CY
We present AI-VERDE, a unified LLM-as-a-platform service designed to facilitate seamless integration of commercial, cloud-hosted, and on-premise open LLMs in academic settings. AI-VERDE streamlines access management for instructional and research groups by providing features such as robust access control, privacy-preserving mechanisms, native Retrieval-Augmented Generation (RAG) support, budget management for third-party LLM services, and both a conversational web interface and API access. In a pilot deployment at a large public university, AI-VERDE demonstrated significant engagement across diverse educational and research groups, enabling activities that would typically require substantial budgets for commercial LLM services with limited user and team management capabilities. To the best of our knowledge, AI-Verde is the first platform to address both academic and research needs for LLMs within an higher education institutional framework.
2502.09652
GraphCompNet: A Position-Aware Model for Predicting and Compensating Shape Deviations in 3D Printing
cs.CV cs.LG
This paper introduces a data-driven algorithm for modeling and compensating shape deviations in additive manufacturing (AM), addressing challenges in geometric accuracy and batch production. While traditional methods, such as analytical models and metrology, laid the groundwork for geometric precision, they are often impractical for large-scale production. Recent advancements in machine learning (ML) have improved compensation precision, but issues remain in generalizing across complex geometries and adapting to position-dependent variations. We present a novel approach for powder bed fusion (PBF) processes, using GraphCompNet, which is a computational framework combining graph-based neural networks with a generative adversarial network (GAN)-inspired training process. By leveraging point cloud data and dynamic graph convolutional neural networks (DGCNNs), GraphCompNet models complex shapes and incorporates position-specific thermal and mechanical factors. A two-stage adversarial training procedure iteratively refines compensated designs via a compensator-predictor architecture, offering real-time feedback and optimization. Experimental validation across diverse shapes and positions shows the framework significantly improves compensation accuracy (35 to 65 percent) across the entire print space, adapting to position-dependent variations. This work advances the development of Digital Twin technology for AM, enabling scalable, real-time monitoring and compensation, and addressing critical gaps in AM process control. The proposed method supports high-precision, automated industrial-scale design and manufacturing systems.
2502.09653
SASVi -- Segment Any Surgical Video
eess.IV cs.CV
Purpose: Foundation models, trained on multitudes of public datasets, often require additional fine-tuning or re-prompting mechanisms to be applied to visually distinct target domains such as surgical videos. Further, without domain knowledge, they cannot model the specific semantics of the target domain. Hence, when applied to surgical video segmentation, they fail to generalise to sections where previously tracked objects leave the scene or new objects enter. Methods: We propose SASVi, a novel re-prompting mechanism based on a frame-wise Mask R-CNN Overseer model, which is trained on a minimal amount of scarcely available annotations for the target domain. This model automatically re-prompts the foundation model SAM2 when the scene constellation changes, allowing for temporally smooth and complete segmentation of full surgical videos. Results: Re-prompting based on our Overseer model significantly improves the temporal consistency of surgical video segmentation compared to similar prompting techniques and especially frame-wise segmentation, which neglects temporal information, by at least 1.5%. Our proposed approach allows us to successfully deploy SAM2 to surgical videos, which we quantitatively and qualitatively demonstrate for three different cholecystectomy and cataract surgery datasets. Conclusion: SASVi can serve as a new baseline for smooth and temporally consistent segmentation of surgical videos with scarcely available annotation data. Our method allows us to leverage scarce annotations and obtain complete annotations for full videos of the large-scale counterpart datasets. We make those annotations publicly available, providing extensive annotation data for the future development of surgical data science models.
2502.09654
Heterogeneous Mixture of Experts for Remote Sensing Image Super-Resolution
eess.IV cs.CV
Remote sensing image super-resolution (SR) aims to reconstruct high-resolution remote sensing images from low-resolution inputs, thereby addressing limitations imposed by sensors and imaging conditions. However, the inherent characteristics of remote sensing images, including diverse ground object types and complex details, pose significant challenges to achieving high-quality reconstruction. Existing methods typically employ a uniform structure to process various types of ground objects without distinction, making it difficult to adapt to the complex characteristics of remote sensing images. To address this issue, we introduce a Mixture of Experts (MoE) model and design a set of heterogeneous experts. These experts are organized into multiple expert groups, where experts within each group are homogeneous while being heterogeneous across groups. This design ensures that specialized activation parameters can be employed to handle the diverse and intricate details of ground objects effectively. To better accommodate the heterogeneous experts, we propose a multi-level feature aggregation strategy to guide the routing process. Additionally, we develop a dual-routing mechanism to adaptively select the optimal expert for each pixel. Experiments conducted on the UCMerced and AID datasets demonstrate that our proposed method achieves superior SR reconstruction accuracy compared to state-of-the-art methods. The code will be available at https://github.com/Mr-Bamboo/MFG-HMoE.
2502.09655
Bidirectional Diffusion Bridge Models
cs.CV cs.AI
Diffusion bridges have shown potential in paired image-to-image (I2I) translation tasks. However, existing methods are limited by their unidirectional nature, requiring separate models for forward and reverse translations. This not only doubles the computational cost but also restricts their practicality. In this work, we introduce the Bidirectional Diffusion Bridge Model (BDBM), a scalable approach that facilitates bidirectional translation between two coupled distributions using a single network. BDBM leverages the Chapman-Kolmogorov Equation for bridges, enabling it to model data distribution shifts across timesteps in both forward and backward directions by exploiting the interchangeability of the initial and target timesteps within this framework. Notably, when the marginal distribution given endpoints is Gaussian, BDBM's transition kernels in both directions possess analytical forms, allowing for efficient learning with a single network. We demonstrate the connection between BDBM and existing bridge methods, such as Doob's h-transform and variational approaches, and highlight its advantages. Extensive experiments on high-resolution I2I translation tasks demonstrate that BDBM not only enables bidirectional translation with minimal additional cost but also outperforms state-of-the-art bridge models. Our source code is available at [https://github.com/kvmduc/BDBM||https://github.com/kvmduc/BDBM].
2502.09656
Multi-Omics Fusion with Soft Labeling for Enhanced Prediction of Distant Metastasis in Nasopharyngeal Carcinoma Patients after Radiotherapy
q-bio.QM cs.CV eess.IV
Omics fusion has emerged as a crucial preprocessing approach in the field of medical image processing, providing significant assistance to several studies. One of the challenges encountered in the integration of omics data is the presence of unpredictability arising from disparities in data sources and medical imaging equipment. In order to overcome this challenge and facilitate the integration of their joint application to specific medical objectives, this study aims to develop a fusion methodology that mitigates the disparities inherent in omics data. The utilization of the multi-kernel late-fusion method has gained significant popularity as an effective strategy for addressing this particular challenge. An efficient representation of the data may be achieved by utilizing a suitable single-kernel function to map the inherent features and afterward merging them in a space with a high number of dimensions. This approach effectively addresses the differences noted before. The inflexibility of label fitting poses a constraint on the use of multi-kernel late-fusion methods in complex nasopharyngeal carcinoma (NPC) datasets, hence affecting the efficacy of general classifiers in dealing with high-dimensional characteristics. This innovative methodology aims to increase the disparity between the two cohorts, hence providing a more flexible structure for the allocation of labels. The examination of the NPC-ContraParotid dataset demonstrates the model's robustness and efficacy, indicating its potential as a valuable tool for predicting distant metastases in patients with nasopharyngeal carcinoma (NPC).
2502.09657
Integrating Spatiotemporal Vision Transformer into Digital Twins for High-Resolution Heat Stress Forecasting in Campus Environments
cs.CV
Extreme heat events exacerbated by climate change pose significant challenges to urban resilience and planning. This study introduces a climate-responsive digital twin framework integrating the Spatiotemporal Vision Transformer (ST-ViT) model to enhance heat stress forecasting and decision-making. Using a Texas campus as a testbed, we synthesized high-resolution physical model simulations with spatial and meteorological data to develop fine-scale human thermal predictions. The ST-ViT-powered digital twin enables efficient, data-driven insights for planners, policymakers, and campus stakeholders, supporting targeted heat mitigation strategies and advancing climate-adaptive urban design.
2502.09658
Neuro-Conceptual Artificial Intelligence: Integrating OPM with Deep Learning to Enhance Question Answering Quality
cs.CL cs.AI
Knowledge representation and reasoning are critical challenges in Artificial Intelligence (AI), particularly in integrating neural and symbolic approaches to achieve explainable and transparent AI systems. Traditional knowledge representation methods often fall short of capturing complex processes and state changes. We introduce Neuro-Conceptual Artificial Intelligence (NCAI), a specialization of the neuro-symbolic AI approach that integrates conceptual modeling using Object-Process Methodology (OPM) ISO 19450:2024 with deep learning to enhance question-answering (QA) quality. By converting natural language text into OPM models using in-context learning, NCAI leverages the expressive power of OPM to represent complex OPM elements-processes, objects, and states-beyond what traditional triplet-based knowledge graphs can easily capture. This rich structured knowledge representation improves reasoning transparency and answer accuracy in an OPM-QA system. We further propose transparency evaluation metrics to quantitatively measure how faithfully the predicted reasoning aligns with OPM-based conceptual logic. Our experiments demonstrate that NCAI outperforms traditional methods, highlighting its potential for advancing neuro-symbolic AI by providing rich knowledge representations, measurable transparency, and improved reasoning.
2502.09659
Cancer Vaccine Adjuvant Name Recognition from Biomedical Literature using Large Language Models
cs.CL cs.AI cs.CY
Motivation: An adjuvant is a chemical incorporated into vaccines that enhances their efficacy by improving the immune response. Identifying adjuvant names from cancer vaccine studies is essential for furthering research and enhancing immunotherapies. However, the manual curation from the constantly expanding biomedical literature poses significant challenges. This study explores the automated recognition of vaccine adjuvant names using Large Language Models (LLMs), specifically Generative Pretrained Transformers (GPT) and Large Language Model Meta AI (Llama). Methods: We utilized two datasets: 97 clinical trial records from AdjuvareDB and 290 abstracts annotated with the Vaccine Adjuvant Compendium (VAC). GPT-4o and Llama 3.2 were employed in zero-shot and few-shot learning paradigms with up to four examples per prompt. Prompts explicitly targeted adjuvant names, testing the impact of contextual information such as substances or interventions. Outputs underwent automated and manual validation for accuracy and consistency. Results: GPT-4o attained 100% Precision across all situations while exhibiting notable improve in Recall and F1-scores, particularly with incorporating interventions. On the VAC dataset, GPT-4o achieved a maximum F1-score of 77.32% with interventions, surpassing Llama-3.2-3B by approximately 2%. On the AdjuvareDB dataset, GPT-4o reached an F1-score of 81.67% for three-shot prompting with interventions, surpassing Llama-3.2-3 B's maximum F1-score of 65.62%. Conclusion: Our findings demonstrate that LLMs excel at identifying adjuvant names, including rare variations of naming representation. This study emphasizes the capability of LLMs to enhance cancer vaccine development by efficiently extracting insights. Future work aims to broaden the framework to encompass various biomedical literature and enhance model generalizability across various vaccines and adjuvants.
2502.09660
Towards Fine-grained Interactive Segmentation in Images and Videos
cs.CV eess.IV
The recent Segment Anything Models (SAMs) have emerged as foundational visual models for general interactive segmentation. Despite demonstrating robust generalization abilities, they still suffer performance degradations in scenarios demanding accurate masks. Existing methods for high-precision interactive segmentation face a trade-off between the ability to perceive intricate local details and maintaining stable prompting capability, which hinders the applicability and effectiveness of foundational segmentation models. To this end, we present an SAM2Refiner framework built upon the SAM2 backbone. This architecture allows SAM2 to generate fine-grained segmentation masks for both images and videos while preserving its inherent strengths. Specifically, we design a localization augment module, which incorporates local contextual cues to enhance global features via a cross-attention mechanism, thereby exploiting potential detailed patterns and maintaining semantic information. Moreover, to strengthen the prompting ability toward the enhanced object embedding, we introduce a prompt retargeting module to renew the embedding with spatially aligned prompt features. In addition, to obtain accurate high resolution segmentation masks, a mask refinement module is devised by employing a multi-scale cascaded structure to fuse mask features with hierarchical representations from the encoder. Extensive experiments demonstrate the effectiveness of our approach, revealing that the proposed method can produce highly precise masks for both images and videos, surpassing state-of-the-art methods.
2502.09662
Generalizable Cervical Cancer Screening via Large-scale Pretraining and Test-Time Adaptation
q-bio.QM cs.CV eess.IV
Cervical cancer is a leading malignancy in female reproductive system. While AI-assisted cytology offers a cost-effective and non-invasive screening solution, current systems struggle with generalizability in complex clinical scenarios. To address this issue, we introduced Smart-CCS, a generalizable Cervical Cancer Screening paradigm based on pretraining and adaptation to create robust and generalizable screening systems. To develop and validate Smart-CCS, we first curated a large-scale, multi-center dataset named CCS-127K, which comprises a total of 127,471 cervical cytology whole-slide images collected from 48 medical centers. By leveraging large-scale self-supervised pretraining, our CCS models are equipped with strong generalization capability, potentially generalizing across diverse scenarios. Then, we incorporated test-time adaptation to specifically optimize the trained CCS model for complex clinical settings, which adapts and refines predictions, improving real-world applicability. We conducted large-scale system evaluation among various cohorts. In retrospective cohorts, Smart-CCS achieved an overall area under the curve (AUC) value of 0.965 and sensitivity of 0.913 for cancer screening on 11 internal test datasets. In external testing, system performance maintained high at 0.950 AUC across 6 independent test datasets. In prospective cohorts, our Smart-CCS achieved AUCs of 0.947, 0.924, and 0.986 in three prospective centers, respectively. Moreover, the system demonstrated superior sensitivity in diagnosing cervical cancer, confirming the accuracy of our cancer screening results by using histology findings for validation. Interpretability analysis with cell and slide predictions further indicated that the system's decision-making aligns with clinical practice. Smart-CCS represents a significant advancement in cancer screening across diverse clinical contexts.
2502.09663
DiffEx: Explaining a Classifier with Diffusion Models to Identify Microscopic Cellular Variations
cs.CV cs.AI cs.LG q-bio.CB
In recent years, deep learning models have been extensively applied to biological data across various modalities. Discriminative deep learning models have excelled at classifying images into categories (e.g., healthy versus diseased, treated versus untreated). However, these models are often perceived as black boxes due to their complexity and lack of interpretability, limiting their application in real-world biological contexts. In biological research, explainability is essential: understanding classifier decisions and identifying subtle differences between conditions are critical for elucidating the effects of treatments, disease progression, and biological processes. To address this challenge, we propose DiffEx, a method for generating visually interpretable attributes to explain classifiers and identify microscopic cellular variations between different conditions. We demonstrate the effectiveness of DiffEx in explaining classifiers trained on natural and biological images. Furthermore, we use DiffEx to uncover phenotypic differences within microscopy datasets. By offering insights into cellular variations through classifier explanations, DiffEx has the potential to advance the understanding of diseases and aid drug discovery by identifying novel biomarkers.
2502.09664
Image Super-Resolution with Guarantees via Conformal Generative Models
cs.CV cs.LG stat.ML
The increasing use of generative ML foundation models for image super-resolution calls for robust and interpretable uncertainty quantification methods. We address this need by presenting a novel approach based on conformal prediction techniques to create a "confidence mask" capable of reliably and intuitively communicating where the generated image can be trusted. Our method is adaptable to any black-box generative model, including those locked behind an opaque API, requires only easily attainable data for calibration, and is highly customizable via the choice of a local image similarity metric. We prove strong theoretical guarantees for our method that span fidelity error control (according to our local image similarity metric), reconstruction quality, and robustness in the face of data leakage. Finally, we empirically evaluate these results and establish our method's solid performance.
2502.09665
Revealing Subtle Phenotypes in Small Microscopy Datasets Using Latent Diffusion Models
cs.CV
Identifying subtle phenotypic variations in cellular images is critical for advancing biological research and accelerating drug discovery. These variations are often masked by the inherent cellular heterogeneity, making it challenging to distinguish differences between experimental conditions. Recent advancements in deep generative models have demonstrated significant potential for revealing these nuanced phenotypes through image translation, opening new frontiers in cellular and molecular biology as well as the identification of novel biomarkers. Among these generative models, diffusion models stand out for their ability to produce high-quality, realistic images. However, training diffusion models typically requires large datasets and substantial computational resources, both of which can be limited in biological research. In this work, we propose a novel approach that leverages pre-trained latent diffusion models to uncover subtle phenotypic changes. We validate our approach qualitatively and quantitatively on several small datasets of microscopy images. Our findings reveal that our approach enables effective detection of phenotypic variations, capturing both visually apparent and imperceptible differences. Ultimately, our results highlight the promising potential of this approach for phenotype detection, especially in contexts constrained by limited data and computational capacity.
2502.09667
k-LLMmeans: Summaries as Centroids for Interpretable and Scalable LLM-Based Text Clustering
cs.CL cs.LG stat.ML
We introduce k-LLMmeans, a novel modification of the k-means clustering algorithm that utilizes LLMs to generate textual summaries as cluster centroids, thereby capturing contextual and semantic nuances often lost when relying on purely numerical means of document embeddings. This modification preserves the properties of k-means while offering greater interpretability: the cluster centroid is represented by an LLM-generated summary, whose embedding guides cluster assignments. We also propose a mini-batch variant, enabling efficient online clustering for streaming text data and providing real-time interpretability of evolving cluster centroids. Through extensive simulations, we show that our methods outperform vanilla k-means on multiple metrics while incurring only modest LLM usage that does not scale with dataset size. Finally, We present a case study showcasing the interpretability of evolving cluster centroids in sequential text streams. As part of our evaluation, we compile a new dataset from StackExchange, offering a benchmark for text-stream clustering.
2502.09669
Meta-INR: Efficient Encoding of Volumetric Data via Meta-Learning Implicit Neural Representation
cs.CV cs.AI cs.GR
Implicit neural representation (INR) has emerged as a promising solution for encoding volumetric data, offering continuous representations and seamless compatibility with the volume rendering pipeline. However, optimizing an INR network from randomly initialized parameters for each new volume is computationally inefficient, especially for large-scale time-varying or ensemble volumetric datasets where volumes share similar structural patterns but require independent training. To close this gap, we propose Meta-INR, a pretraining strategy adapted from meta-learning algorithms to learn initial INR parameters from partial observation of a volumetric dataset. Compared to training an INR from scratch, the learned initial parameters provide a strong prior that enhances INR generalizability, allowing significantly faster convergence with just a few gradient updates when adapting to a new volume and better interpretability when analyzing the parameters of the adapted INRs. We demonstrate that Meta-INR can effectively extract high-quality generalizable features that help encode unseen similar volume data across diverse datasets. Furthermore, we highlight its utility in tasks such as simulation parameter analysis and representative timestep selection. The code is available at https://github.com/spacefarers/MetaINR.
2502.09670
The Science of Evaluating Foundation Models
cs.CL cs.AI
The emergent phenomena of large foundation models have revolutionized natural language processing. However, evaluating these models presents significant challenges due to their size, capabilities, and deployment across diverse applications. Existing literature often focuses on individual aspects, such as benchmark performance or specific tasks, but fails to provide a cohesive process that integrates the nuances of diverse use cases with broader ethical and operational considerations. This work focuses on three key aspects: (1) Formalizing the Evaluation Process by providing a structured framework tailored to specific use-case contexts, (2) Offering Actionable Tools and Frameworks such as checklists and templates to ensure thorough, reproducible, and practical evaluations, and (3) Surveying Recent Work with a targeted review of advancements in LLM evaluation, emphasizing real-world applications.
2502.09672
IMM-MOT: A Novel 3D Multi-object Tracking Framework with Interacting Multiple Model Filter
cs.CV cs.RO
3D Multi-Object Tracking (MOT) provides the trajectories of surrounding objects, assisting robots or vehicles in smarter path planning and obstacle avoidance. Existing 3D MOT methods based on the Tracking-by-Detection framework typically use a single motion model to track an object throughout its entire tracking process. However, objects may change their motion patterns due to variations in the surrounding environment. In this paper, we introduce the Interacting Multiple Model filter in IMM-MOT, which accurately fits the complex motion patterns of individual objects, overcoming the limitation of single-model tracking in existing approaches. In addition, we incorporate a Damping Window mechanism into the trajectory lifecycle management, leveraging the continuous association status of trajectories to control their creation and termination, reducing the occurrence of overlooked low-confidence true targets. Furthermore, we propose the Distance-Based Score Enhancement module, which enhances the differentiation between false positives and true positives by adjusting detection scores, thereby improving the effectiveness of the Score Filter. On the NuScenes Val dataset, IMM-MOT outperforms most other single-modal models using 3D point clouds, achieving an AMOTA of 73.8%. Our project is available at https://github.com/Ap01lo/IMM-MOT.
2502.09673
Are Smarter LLMs Safer? Exploring Safety-Reasoning Trade-offs in Prompting and Fine-Tuning
cs.CL cs.AI
Large Language Models (LLMs) have demonstrated remarkable success across various NLP benchmarks. However, excelling in complex tasks that require nuanced reasoning and precise decision-making demands more than raw language proficiency--LLMs must reason, i.e., think logically, draw from past experiences, and synthesize information to reach conclusions and take action. To enhance reasoning abilities, approaches such as prompting and fine-tuning have been widely explored. While these methods have led to clear improvements in reasoning, their impact on LLM safety remains less understood. In this work, we investigate the interplay between reasoning and safety in LLMs. We highlight the latent safety risks that arise as reasoning capabilities improve, shedding light on previously overlooked vulnerabilities. At the same time, we explore how reasoning itself can be leveraged to enhance safety, uncovering potential mitigation strategies. By examining both the risks and opportunities in reasoning-driven LLM safety, our study provides valuable insights for developing models that are not only more capable but also more trustworthy in real-world deployments.
2502.09674
The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Safety Analysis
cs.CL cs.AI
Large Language Models' safety-aligned behaviors, such as refusing harmful queries, can be represented by linear directions in activation space. Previous research modeled safety behavior with a single direction, limiting mechanistic understanding to an isolated safety feature. In this work, we discover that safety-aligned behavior is jointly controlled by multi-dimensional directions. Namely, we study the vector space of representation shifts during safety fine-tuning on Llama 3 8B for refusing jailbreaks. By studying orthogonal directions in the space, we first find that a dominant direction governs the model's refusal behavior, while multiple smaller directions represent distinct and interpretable features like hypothetical narrative and role-playing. We then measure how different directions promote or suppress the dominant direction, showing the important role of secondary directions in shaping the model's refusal representation. Finally, we demonstrate that removing certain trigger tokens in harmful queries can mitigate these directions to bypass the learned safety capability, providing new insights on understanding safety alignment vulnerability from a multi-dimensional perspective. Code and artifacts are available at https://github.com/BMPixel/safety-residual-space.
2502.09675
Multi-level Conflict-Aware Network for Multi-modal Sentiment Analysis
cs.CL cs.AI cs.LG
Multimodal Sentiment Analysis (MSA) aims to recognize human emotions by exploiting textual, acoustic, and visual modalities, and thus how to make full use of the interactions between different modalities is a central challenge of MSA. Interaction contains alignment and conflict aspects. Current works mainly emphasize alignment and the inherent differences between unimodal modalities, neglecting the fact that there are also potential conflicts between bimodal combinations. Additionally, multi-task learning-based conflict modeling methods often rely on the unstable generated labels. To address these challenges, we propose a novel multi-level conflict-aware network (MCAN) for multimodal sentiment analysis, which progressively segregates alignment and conflict constituents from unimodal and bimodal representations, and further exploits the conflict constituents with the conflict modeling branch. In the conflict modeling branch, we conduct discrepancy constraints at both the representation and predicted output levels, avoiding dependence on the generated labels. Experimental results on the CMU-MOSI and CMU-MOSEI datasets demonstrate the effectiveness of the proposed MCAN.
2502.09680
Object-Centric Latent Action Learning
cs.CV cs.AI
Leveraging vast amounts of internet video data for Embodied AI is currently bottle-necked by the lack of action annotations and the presence of action-correlated distractors. We propose a novel object-centric latent action learning approach, based on VideoSaur and LAPO, that employs self-supervised decomposition of scenes into object representations and annotates video data with proxy-action labels. This method effectively disentangles causal agent-object interactions from irrelevant background noise and reduces the performance degradation of latent action learning approaches caused by distractors. Our preliminary experiments with the Distracting Control Suite show that latent action pretraining based on object decompositions improve the quality of inferred latent actions by x2.7 and efficiency of downstream fine-tuning with a small set of labeled actions, increasing return by x2.6 on average.
2502.09682
Lifespan tree of brain anatomy: diagnostic values for motor and cognitive neurodegenerative diseases
eess.IV cs.LG
The differential diagnosis of neurodegenerative diseases, characterized by overlapping symptoms, may be challenging. Brain imaging coupled with artificial intelligence has been previously proposed for diagnostic support, but most of these methods have been trained to discriminate only isolated diseases from controls. Here, we develop a novel machine learning framework, named lifespan tree of brain anatomy, dedicated to the differential diagnosis between multiple diseases simultaneously. It integrates the modeling of volume changes for 124 brain structures during the lifespan with non-linear dimensionality reduction and synthetic sampling techniques to create easily interpretable representations of brain anatomy over the course of disease progression. As clinically relevant proof-of-concept applications, we constructed a cognitive lifespan tree of brain anatomy for the differential diagnosis of six causes of neurodegenerative dementia and a motor lifespan tree of brain anatomy for the differential diagnosis of four causes of parkinsonism using 37594 MRI as a training dataset. This original approach enhanced significantly the efficiency of differential diagnosis in the external validation cohort of 1754 cases, outperforming existing state-of-the art machine learning techniques. Lifespan tree holds promise as a valuable tool for differential diagnostic in relevant clinical conditions, especially for diseases still lacking effective biological markers.
2502.09683
Channel Dependence, Limited Lookback Windows, and the Simplicity of Datasets: How Biased is Time Series Forecasting?
cs.LG
Time-series forecasting research has converged to a small set of datasets and a standardized collection of evaluation scenarios. Such a standardization is to a specific extent needed for comparable research. However, the underlying assumption is, that the considered setting is a representative for the problem as a whole. In this paper, we challenge this assumption and show that the current scenario gives a strongly biased perspective on the state of time-series forecasting research. To be more detailed, we show that the current evaluation scenario is heavily biased by the simplicity of the current datasets. We furthermore emphasize, that when the lookback-window is properly tuned, current models usually do not need any information flow across channels. However, when using more complex benchmark data, the situation changes: Here, modeling channel-interactions in a sophisticated manner indeed enhances performances. Furthermore, in this complex evaluation scenario, Crossformer, a method regularly neglected as an important baseline, is the SOTA method for time series forecasting. Based on this, we present the Fast Channel-dependent Transformer (FaCT), a simplified version of Crossformer which closes the runtime gap between Crossformer and TimeMixer, leading to an efficient model for complex forecasting datasets.
2502.09685
A Novel Hybrid Approach to Contraceptive Demand Forecasting: Integrating Point Predictions with Probabilistic Distributions
cs.LG stat.AP stat.ME
Accurate demand forecasting is vital for ensuring reliable access to contraceptive products, supporting key processes like procurement, inventory, and distribution. However, forecasting contraceptive demand in developing countries presents challenges, including incomplete data, poor data quality, and the need to account for multiple geographical and product factors. Current methods often rely on simple forecasting techniques, which fail to capture demand uncertainties arising from these factors, warranting expert involvement. Our study aims to improve contraceptive demand forecasting by combining probabilistic forecasting methods with expert knowledge. We developed a hybrid model that combines point forecasts from domain-specific model with probabilistic distributions from statistical and machine learning approaches, enabling human input to fine-tune and enhance the system-generated forecasts. This approach helps address the uncertainties in demand and is particularly useful in resource-limited settings. We evaluate different forecasting methods, including time series, Bayesian, machine learning, and foundational time series methods alongside our new hybrid approach. By comparing these methods, we provide insights into their strengths, weaknesses, and computational requirements. Our research fills a gap in forecasting contraceptive demand and offers a practical framework that combines algorithmic and human expertise. Our proposed model can also be generalized to other humanitarian contexts with similar data patterns.
2502.09686
Leveraging Machine Learning and Deep Learning Techniques for Improved Pathological Staging of Prostate Cancer
cs.LG
Prostate cancer (Pca) continues to be a leading cause of cancer-related mortality in men, and the limitations in precision of traditional diagnostic methods such as the Digital Rectal Exam (DRE), Prostate-Specific Antigen (PSA) testing, and biopsies underscore the critical importance of accurate staging detection in enhancing treatment outcomes and improving patient prognosis. This study leverages machine learning and deep learning approaches, along with feature selection and extraction methods, to enhance PCa pathological staging predictions using RNA sequencing data from The Cancer Genome Atlas (TCGA). Gene expression profiles from 486 tumors were analyzed using advanced algorithms, including Random Forest (RF), Logistic Regression (LR), Extreme Gradient Boosting (XGB), and Support Vector Machine (SVM). The performance of the study is measured with respect to the F1-score, as well as precision and recall, all of which are calculated as weighted averages. The results reveal that the highest test F1-score, approximately 83%, was achieved by the Random Forest algorithm, followed by Logistic Regression at 80%, while both Extreme Gradient Boosting (XGB) and Support Vector Machine (SVM) scored around 79%. Furthermore, deep learning models with data augmentation achieved an accuracy of 71. 23%, while PCA-based dimensionality reduction reached an accuracy of 69.86%. This research highlights the potential of AI-driven approaches in clinical oncology, paving the way for more reliable diagnostic tools that can ultimately improve patient outcomes.
2502.09687
Mind What You Ask For: Emotional and Rational Faces of Persuasion by Large Language Models
cs.CL cs.AI cs.HC
Be careful what you ask for, you just might get it. This saying fits with the way large language models (LLMs) are trained, which, instead of being rewarded for correctness, are increasingly rewarded for pleasing the recipient. So, they are increasingly effective at persuading us that their answers are valuable. But what tricks do they use in this persuasion? In this study, we examine what are the psycholinguistic features of the responses used by twelve different language models. By grouping response content according to rational or emotional prompts and exploring social influence principles employed by LLMs, we ask whether and how we can mitigate the risks of LLM-driven mass misinformation. We position this study within the broader discourse on human-centred AI, emphasizing the need for interdisciplinary approaches to mitigate cognitive and societal risks posed by persuasive AI responses.
2502.09688
Towards Virtual Clinical Trials of Radiology AI with Conditional Generative Modeling
cs.CV cs.AI cs.LG
Artificial intelligence (AI) is poised to transform healthcare by enabling personalized and efficient care through data-driven insights. Although radiology is at the forefront of AI adoption, in practice, the potential of AI models is often overshadowed by severe failures to generalize: AI models can have performance degradation of up to 20% when transitioning from controlled test environments to clinical use by radiologists. This mismatch raises concerns that radiologists will be misled by incorrect AI predictions in practice and/or grow to distrust AI, rendering these promising technologies practically ineffectual. Exhaustive clinical trials of AI models on abundant and diverse data is thus critical to anticipate AI model degradation when encountering varied data samples. Achieving these goals, however, is challenging due to the high costs of collecting diverse data samples and corresponding annotations. To overcome these limitations, we introduce a novel conditional generative AI model designed for virtual clinical trials (VCTs) of radiology AI, capable of realistically synthesizing full-body CT images of patients with specified attributes. By learning the joint distribution of images and anatomical structures, our model enables precise replication of real-world patient populations with unprecedented detail at this scale. We demonstrate meaningful evaluation of radiology AI models through VCTs powered by our synthetic CT study populations, revealing model degradation and facilitating algorithmic auditing for bias-inducing data attributes. Our generative AI approach to VCTs is a promising avenue towards a scalable solution to assess model robustness, mitigate biases, and safeguard patient care by enabling simpler testing and evaluation of AI models in any desired range of diverse patient populations.
2502.09689
Large Language Models and Provenance Metadata for Determining the Relevance of Images and Videos in News Stories
cs.CL cs.CV cs.CY
The most effective misinformation campaigns are multimodal, often combining text with images and videos taken out of context -- or fabricating them entirely -- to support a given narrative. Contemporary methods for detecting misinformation, whether in deepfakes or text articles, often miss the interplay between multiple modalities. Built around a large language model, the system proposed in this paper addresses these challenges. It analyzes both the article's text and the provenance metadata of included images and videos to determine whether they are relevant. We open-source the system prototype and interactive web interface.
2502.09690
Trust at Your Own Peril: A Mixed Methods Exploration of the Ability of Large Language Models to Generate Expert-Like Systems Engineering Artifacts and a Characterization of Failure Modes
cs.CL cs.AI
Multi-purpose Large Language Models (LLMs), a subset of generative Artificial Intelligence (AI), have recently made significant progress. While expectations for LLMs to assist systems engineering (SE) tasks are paramount; the interdisciplinary and complex nature of systems, along with the need to synthesize deep-domain knowledge and operational context, raise questions regarding the efficacy of LLMs to generate SE artifacts, particularly given that they are trained using data that is broadly available on the internet. To that end, we present results from an empirical exploration, where a human expert-generated SE artifact was taken as a benchmark, parsed, and fed into various LLMs through prompt engineering to generate segments of typical SE artifacts. This procedure was applied without any fine-tuning or calibration to document baseline LLM performance. We then adopted a two-fold mixed-methods approach to compare AI generated artifacts against the benchmark. First, we quantitatively compare the artifacts using natural language processing algorithms and find that when prompted carefully, the state-of-the-art algorithms cannot differentiate AI-generated artifacts from the human-expert benchmark. Second, we conduct a qualitative deep dive to investigate how they differ in terms of quality. We document that while the two-material appear very similar, AI generated artifacts exhibit serious failure modes that could be difficult to detect. We characterize these as: premature requirements definition, unsubstantiated numerical estimates, and propensity to overspecify. We contend that this study tells a cautionary tale about why the SE community must be more cautious adopting AI suggested feedback, at least when generated by multi-purpose LLMs.
2502.09692
NeuralCFD: Deep Learning on High-Fidelity Automotive Aerodynamics Simulations
cs.LG cs.AI
Recent advancements in neural operator learning are paving the way for transformative innovations in fields such as automotive aerodynamics. However, key challenges must be overcome before neural network-based simulation surrogates can be implemented at an industry scale. First, surrogates must become scalable to large surface and volume meshes, especially when using raw geometry inputs only, i.e., without relying on the simulation mesh. Second, surrogates must be trainable with a limited number of high-fidelity numerical simulation samples while still reaching the required performance levels. To this end, we introduce Geometry-preserving Universal Physics Transformer (GP-UPT), which separates geometry encoding and physics predictions, ensuring flexibility with respect to geometry representations and surface sampling strategies. GP-UPT enables independent scaling of the respective parts of the model according to practical requirements, offering scalable solutions to open challenges. GP-UPT circumvents the creation of high-quality simulation meshes, enables accurate 3D velocity field predictions at 20 million mesh cells, and excels in transfer learning from low-fidelity to high-fidelity simulation datasets, requiring less than half of the high-fidelity data to match the performance of models trained from scratch.
2502.09695
Power System Electromagnetic Transient Stability: an Analysis Based on Convergent Hamiltonian
eess.SY cs.SY
Transient stability is crucial to the reliable operation of power systems. Existing theories rely on the simplified electromechanical models, substituting the detailed electromagnetic dynamics of inductor and capacitor with their impedance representations. However, this simplification is inadequate for the growing penetration of fast-switching power electronic devices. Attempts to extend the existing theories to include electromagnetic dynamics lead to overly conservative stability conditions. To tackle this problem more directly, we study the condition under which the power source and dissipation in the electromagnetic dynamics tend to balance each other asymptotically. This is equivalent to the convergence of the Hamiltonian (total stored energy) and can be shown to imply transient stability. Using contraction analysis, we prove that this property holds for a large class of time-varying port-Hamiltonian systems with (i) constant damping matrix and (ii) strictly convex Hamiltonian. Then through port-Hamiltonian modeling of the electromagnetic dynamics, we obtain that the synchronized steady state of the power system is globally stable if it exists. This result provides new insights into the reliable operation of power systems. The proposed theory is illustrated in the simulation results of a two-machine system.
2502.09696
ZeroBench: An Impossible Visual Benchmark for Contemporary Large Multimodal Models
cs.CV
Large Multimodal Models (LMMs) exhibit major shortfalls when interpreting images and, by some measures, have poorer spatial cognition than small children or animals. Despite this, they attain high scores on many popular visual benchmarks, with headroom rapidly eroded by an ongoing surge of model progress. To address this, there is a pressing need for difficult benchmarks that remain relevant for longer. We take this idea to its limit by introducing ZeroBench-a lightweight visual reasoning benchmark that is entirely impossible for contemporary frontier LMMs. Our benchmark consists of 100 manually curated questions and 334 less difficult subquestions. We evaluate 20 LMMs on ZeroBench, all of which score 0.0%, and rigorously analyse the errors. To encourage progress in visual understanding, we publicly release ZeroBench.
2502.09704
Iterative quantum optimisation with a warm-started quantum state
quant-ph cond-mat.dis-nn cs.LG math.OC physics.comp-ph
We provide a method to prepare a warm-started quantum state from measurements with an iterative framework to enhance the quantum approximate optimisation algorithm (QAOA). The numerical simulations show the method can effectively address the "stuck issue" of the standard QAOA using a single-string warm-started initial state described in [Cain et al., 2023]. When applied to the $3$-regular MaxCut problem, our approach achieves an improved approximation ratio, with a lower bound that iteratively converges toward the best classical algorithms for $p=1$ standard QAOA. Additionally, in the context of the discrete global minimal variance portfolio (DGMVP) model, simulations reveal a more favourable scaling of identifying the global minimal compared to the QAOA standalone, the single-string warm-started QAOA and a classical constrained sampling approach.
2502.09715
Evaluating GPT's Capability in Identifying Stages of Cognitive Impairment from Electronic Health Data
cs.LG cs.AI cs.CL
Identifying cognitive impairment within electronic health records (EHRs) is crucial not only for timely diagnoses but also for facilitating research. Information about cognitive impairment often exists within unstructured clinician notes in EHRs, but manual chart reviews are both time-consuming and error-prone. To address this issue, our study evaluates an automated approach using zero-shot GPT-4o to determine stage of cognitive impairment in two different tasks. First, we evaluated the ability of GPT-4o to determine the global Clinical Dementia Rating (CDR) on specialist notes from 769 patients who visited the memory clinic at Massachusetts General Hospital (MGH), and achieved a weighted kappa score of 0.83. Second, we assessed GPT-4o's ability to differentiate between normal cognition, mild cognitive impairment (MCI), and dementia on all notes in a 3-year window from 860 Medicare patients. GPT-4o attained a weighted kappa score of 0.91 in comparison to specialist chart reviews and 0.96 on cases that the clinical adjudicators rated with high confidence. Our findings demonstrate GPT-4o's potential as a scalable chart review tool for creating research datasets and assisting diagnosis in clinical settings in the future.
2502.09717
Carbon- and Precedence-Aware Scheduling for Data Processing Clusters
cs.DC cs.CY cs.SY eess.SY
As large-scale data processing workloads continue to grow, their carbon footprint raises concerns. Prior research on carbon-aware schedulers has focused on shifting computation to align with availability of low-carbon energy, but these approaches assume that each task can be executed independently. In contrast, data processing jobs have precedence constraints (i.e., outputs of one task are inputs for another) that complicate decisions, since delaying an upstream ``bottleneck'' task to a low-carbon period will also block downstream tasks, impacting the entire job's completion time. In this paper, we show that carbon-aware scheduling for data processing benefits from knowledge of both time-varying carbon and precedence constraints. Our main contribution is $\texttt{PCAPS}$, a carbon-aware scheduler that interfaces with modern ML scheduling policies to explicitly consider the precedence-driven importance of each task in addition to carbon. To illustrate the gains due to fine-grained task information, we also study $\texttt{CAP}$, a wrapper for any carbon-agnostic scheduler that adapts the key provisioning ideas of $\texttt{PCAPS}$. Our schedulers enable a configurable priority between carbon reduction and job completion time, and we give analytical results characterizing the trade-off between the two. Furthermore, our Spark prototype on a 100-node Kubernetes cluster shows that a moderate configuration of $\texttt{PCAPS}$ reduces carbon footprint by up to 32.9% without significantly impacting the cluster's total efficiency.
2502.09720
NestQuant: Nested Lattice Quantization for Matrix Products and LLMs
cs.LG cs.AI cs.IT math.IT
Post-training quantization (PTQ) has emerged as a critical technique for efficient deployment of large language models (LLMs). This work proposes NestQuant, a novel PTQ scheme for weights and activations that is based on self-similar nested lattices. Recent work have mathematically shown such quantizers to be information-theoretically optimal for low-precision matrix multiplication. We implement a practical low-complexity version of NestQuant based on Gosset lattice, making it a drop-in quantizer for any matrix multiplication step (e.g., in self-attention, MLP etc). For example, NestQuant quantizes weights, KV-cache, and activations of Llama-3-8B to 4 bits, achieving perplexity of 6.6 on wikitext2. This represents more than 55% reduction in perplexity gap with respect to unquantized model (perplexity of 6.14) compared to state-of-the-art Meta's SpinQuant (perplexity 7.3). Comparisons on various LLM evaluation benchmarks also show a reduction in performance degradation induced by quantization.
2502.09723
Making Them a Malicious Database: Exploiting Query Code to Jailbreak Aligned Large Language Models
cs.CR cs.AI cs.CL
Recent advances in large language models (LLMs) have demonstrated remarkable potential in the field of natural language processing. Unfortunately, LLMs face significant security and ethical risks. Although techniques such as safety alignment are developed for defense, prior researches reveal the possibility of bypassing such defenses through well-designed jailbreak attacks. In this paper, we propose QueryAttack, a novel framework to examine the generalizability of safety alignment. By treating LLMs as knowledge databases, we translate malicious queries in natural language into structured non-natural query language to bypass the safety alignment mechanisms of LLMs. We conduct extensive experiments on mainstream LLMs, and the results show that QueryAttack not only can achieve high attack success rates (ASRs), but also can jailbreak various defense methods. Furthermore, we tailor a defense method against QueryAttack, which can reduce ASR by up to 64% on GPT-4-1106. Our code is available at https://github.com/horizonsinzqs/QueryAttack.
2502.09724
Navigating the Social Welfare Frontier: Portfolios for Multi-objective Reinforcement Learning
cs.LG
In many real-world applications of reinforcement learning (RL), deployed policies have varied impacts on different stakeholders, creating challenges in reaching consensus on how to effectively aggregate their preferences. Generalized $p$-means form a widely used class of social welfare functions for this purpose, with broad applications in fair resource allocation, AI alignment, and decision-making. This class includes well-known welfare functions such as Egalitarian, Nash, and Utilitarian welfare. However, selecting the appropriate social welfare function is challenging for decision-makers, as the structure and outcomes of optimal policies can be highly sensitive to the choice of $p$. To address this challenge, we study the concept of an $\alpha$-approximate portfolio in RL, a set of policies that are approximately optimal across the family of generalized $p$-means for all $p \in [-\infty, 1]$. We propose algorithms to compute such portfolios and provide theoretical guarantees on the trade-offs among approximation factor, portfolio size, and computational efficiency. Experimental results on synthetic and real-world datasets demonstrate the effectiveness of our approach in summarizing the policy space induced by varying $p$ values, empowering decision-makers to navigate this landscape more effectively.
2502.09728
Perch like a bird: bio-inspired optimal maneuvers and nonlinear control for Flapping-Wing Unmanned Aerial Vehicles
eess.SY cs.RO cs.SY math.OC
This research endeavors to design the perching maneuver and control in ornithopter robots. By analyzing the dynamic interplay between the robot's flight dynamics, feedback loops, and the environmental constraints, we aim to advance our understanding of the perching maneuver, drawing parallels to biological systems. Inspired by the elegant control strategies observed in avian flight, we develop an optimal maneuver and a corresponding controller to achieve stable perching. The maneuver consists of a deceleration and a rapid pitch-up (vertical turn), which arises from analytically solving the optimization problem of minimal velocity at perch, subject to kinematic and dynamic constraints. The controller for the flapping frequency and tail symmetric deflection is nonlinear and adaptive, ensuring robustly stable perching. Indeed, such adaptive behavior in a sense incorporates homeostatic principles of cybernetics into the control system, enhancing the robot's ability to adapt to unexpected disturbances and maintain a stable posture during the perching maneuver. The resulting autonomous perching maneuvers -- closed-loop descent and turn -- , have been verified and validated, demonstrating excellent agreement with real bird perching trajectories reported in the literature. These findings lay the theoretical groundwork for the development of future prototypes that better imitate the skillful perching maneuvers of birds.
2502.09731
A CNN Approach to Automated Detection and Classification of Brain Tumors
cs.CV cs.AI
Brain tumors require an assessment to ensure timely diagnosis and effective patient treatment. Morphological factors such as size, location, texture, and variable appearance complicate tumor inspection. Medical imaging presents challenges, including noise and incomplete images. This research article presents a methodology for processing Magnetic Resonance Imaging (MRI) data, encompassing techniques for image classification and denoising. The effective use of MRI images allows medical professionals to detect brain disorders, including tumors. This research aims to categorize healthy brain tissue and brain tumors by analyzing the provided MRI data. Unlike alternative methods like Computed Tomography (CT), MRI technology offers a more detailed representation of internal anatomical components, making it a suitable option for studying data related to brain tumors. The MRI picture is first subjected to a denoising technique utilizing an Anisotropic diffusion filter. The dataset utilized for the models creation is a publicly accessible and validated Brain Tumour Classification (MRI) database, comprising 3,264 brain MRI scans. SMOTE was employed for data augmentation and dataset balancing. Convolutional Neural Networks(CNN) such as ResNet152V2, VGG, ViT, and EfficientNet were employed for the classification procedure. EfficientNet attained an accuracy of 98%, the highest recorded.
2502.09741
FoNE: Precise Single-Token Number Embeddings via Fourier Features
cs.CL cs.LG
Large Language Models (LLMs) typically represent numbers using multiple tokens, which requires the model to aggregate these tokens to interpret numerical values. This fragmentation makes both training and inference less efficient and adversely affects the model's performance on number-related tasks. Inspired by the observation that pre-trained LLMs internally learn Fourier-like features for number tokens, we propose Fourier Number Embedding (FoNE), a novel method that directly maps numbers into the embedding space with their Fourier features. FoNE encodes each number as a single token with only two embedding dimensions per digit, effectively capturing numerical values without fragmentation. This compact representation accelerates both training and inference. Compared to traditional subword and digit-wise embeddings, FoNE not only reduces computational overhead but also achieves higher accuracy across various numerical tasks including addition, subtraction and multiplication. On 6-digit decimal addition, FoNE requires 64$\times$ less data to achieve 99% accuracy than subword and digit-wise embeddings while using 3$\times$ and 6$\times$ fewer tokens per number, respectively. Furthermore, FoNE is the only method that yields 100% accuracy on over 100,000 test examples for addition, subtraction, and multiplication. The codes and visualization are available at https://fouriernumber.github.io/.
2502.09743
Partial Colexifications Improve Concept Embeddings
cs.CL
While the embedding of words has revolutionized the field of Natural Language Processing, the embedding of concepts has received much less attention so far. A dense and meaningful representation of concepts, however, could prove useful for several tasks in computational linguistics, especially those involving cross-linguistic data or sparse data from low resource languages. First methods that have been proposed so far embed concepts from automatically constructed colexification networks. While these approaches depart from automatically inferred polysemies, attested across a larger number of languages, they are restricted to the word level, ignoring lexical relations that would only hold for parts of the words in a given language. Building on recently introduced methods for the inference of partial colexifications, we show how they can be used to improve concept embeddings in meaningful ways. The learned embeddings are evaluated against lexical similarity ratings, recorded instances of semantic shift, and word association data. We show that in all evaluation tasks, the inclusion of partial colexifications lead to improved concept representations and better results. Our results further show that the learned embeddings are able to capture and represent different semantic relationships between concepts.
2502.09744
Fine-Tuning Foundation Models with Federated Learning for Privacy Preserving Medical Time Series Forecasting
cs.LG cs.CR
Federated Learning (FL) provides a decentralized machine learning approach, where multiple devices or servers collaboratively train a model without sharing their raw data, thus enabling data privacy. This approach has gained significant interest in academia and industry due to its privacy-preserving properties, which are particularly valuable in the medical domain where data availability is often protected under strict regulations. A relatively unexplored area is the use of FL to fine-tune Foundation Models (FMs) for time series forecasting, potentially enhancing model efficacy by overcoming data limitation while maintaining privacy. In this paper, we fine-tuned time series FMs with Electrocardiogram (ECG) and Impedance Cardiography (ICG) data using different FL techniques. We then examined various scenarios and discussed the challenges FL faces under different data heterogeneity configurations. Our empirical results demonstrated that while FL can be effective for fine-tuning FMs on time series forecasting tasks, its benefits depend on the data distribution across clients. We highlighted the trade-offs in applying FL to FM fine-tuning.
2502.09747
The Widespread Adoption of Large Language Model-Assisted Writing Across Society
cs.CL
The recent advances in large language models (LLMs) attracted significant public and policymaker interest in its adoption patterns. In this paper, we systematically analyze LLM-assisted writing across four domains-consumer complaints, corporate communications, job postings, and international organization press releases-from January 2022 to September 2024. Our dataset includes 687,241 consumer complaints, 537,413 corporate press releases, 304.3 million job postings, and 15,919 United Nations (UN) press releases. Using a robust population-level statistical framework, we find that LLM usage surged following the release of ChatGPT in November 2022. By late 2024, roughly 18% of financial consumer complaint text appears to be LLM-assisted, with adoption patterns spread broadly across regions and slightly higher in urban areas. For corporate press releases, up to 24% of the text is attributable to LLMs. In job postings, LLM-assisted writing accounts for just below 10% in small firms, and is even more common among younger firms. UN press releases also reflect this trend, with nearly 14% of content being generated or modified by LLMs. Although adoption climbed rapidly post-ChatGPT, growth appears to have stabilized by 2024, reflecting either saturation in LLM adoption or increasing subtlety of more advanced models. Our study shows the emergence of a new reality in which firms, consumers and even international organizations substantially rely on generative AI for communications.
2502.09748
Contracting Strategies for Electrolyzers to Secure Grid Connection: The Dutch Case
math.OC cs.SY eess.SY
In response to increasing grid congestion in the Netherlands, non-firm connection and transport agreements (CTAs) and capacity restriction contracts (CRCs) have been introduced, allowing consumer curtailment in exchange for grid tariff discounts or per-MW compensations. This study examines the interaction between an electrolyzer project, facing sizing and contracting decisions, and a network operator, responsible for contract activations and determining grid connection capacity, under the new Dutch regulations. The interaction is modeled using two bilevel optimization problems with alternating leader-follower roles. Results highlight a trade-off between CRC income and non-firm CTA tariff discounts, showing that voluntary congestion management by the network operator increases electrolyzer profitability at CRC prices below 10 euro per MW but reduces it at higher prices. Furthermore, the network operator benefits more from reacting to the electrolyzer owner's CTA decisions than from leading the interaction at CRC prices above 10 euro per MW. Ignoring the other party's optimization problem overestimates profits for both the network operator and the electrolyzer owner, emphasizing the importance of coordinated decision-making.
2502.09749
Vote-Tree-Planner: Optimizing Execution Order in LLM-based Task Planning Pipeline via Voting
cs.RO cs.AI
Integrating large language models (LLMs) into closed-loop robotic task planning has become increasingly popular within embodied artificial intelligence. Previous efforts mainly focused on leveraging the strong reasoning abilities of LLMs to enhance task planning performance while often overlooking task planning efficiency and executability due to repetitive queries to LLMs. This paper addresses the synergy between LLMs and task planning systems, aiming to minimize redundancy while enhancing planning effectiveness. Specifically, building upon Prog-Prompt and the high-level concept of Tree-Planner, we propose Vote-Tree-Planner. This sampling strategy utilizes votes to guide plan traversal during the decision-making process. Our approach is motivated by a straightforward observation: assigning weights to agents during decision-making enables the evaluation of critical paths before execution. With this simple vote-tree construction, our method further improves the success rate and reduces the number of queries to LLMs. The experimental results highlight that our Vote-Tree-Planner demonstrates greater stability and shows a higher average success rate and goal condition recall on the unseen dataset compared with previous baseline methods. These findings underscore the potential of the Vote-Tree-Planner to enhance planning accuracy, reliability, and efficiency in LLM-based planning systems.
2502.09755
Enhancing Jailbreak Attacks via Compliance-Refusal-Based Initialization
cs.CR cs.LG
Jailbreak attacks aim to exploit large language models (LLMs) and pose a significant threat to their proper conduct; they seek to bypass models' safeguards and often provoke transgressive behaviors. However, existing automatic jailbreak attacks require extensive computational resources and are prone to converge on suboptimal solutions. In this work, we propose \textbf{C}ompliance \textbf{R}efusal \textbf{I}nitialization (CRI), a novel, attack-agnostic framework that efficiently initializes the optimization in the proximity of the compliance subspace of harmful prompts. By narrowing the initial gap to the adversarial objective, CRI substantially improves adversarial success rates (ASR) and drastically reduces computational overhead -- often requiring just a single optimization step. We evaluate CRI on the widely-used AdvBench dataset over the standard jailbreak attacks of GCG and AutoDAN. Results show that CRI boosts ASR and decreases the median steps to success by up to \textbf{\(\times 60\)}. The project page, along with the reference implementation, is publicly available at \texttt{https://amit1221levi.github.io/CRI-Jailbreak-Init-LLMs-evaluation/}.
2502.09757
The AI-Therapist Duo: Exploring the Potential of Human-AI Collaboration in Personalized Art Therapy for PICS Intervention
cs.HC cs.AI
Post-intensive care syndrome (PICS) is a multifaceted condition that arises from prolonged stays in an intensive care unit (ICU). While preventing PICS among ICU patients is becoming increasingly important, interventions remain limited. Building on evidence supporting the effectiveness of art exposure in addressing the psychological aspects of PICS, we propose a novel art therapy solution through a collaborative Human-AI approach that enhances personalized therapeutic interventions using state-of-the-art Visual Art Recommendation Systems. We developed two Human-in-the-Loop (HITL) personalization methods and assessed their impact through a large-scale user study (N=150). Our findings demonstrate that this Human-AI collaboration not only enhances the personalization and effectiveness of art therapy but also supports therapists by streamlining their workload. While our study centres on PICS intervention, the results suggest that human-AI collaborative Art therapy could potentially benefit other areas where emotional support is critical, such as cases of anxiety and depression.
2502.09762
Adaptive Teaming in Multi-Drone Pursuit: Simulation, Training, and Deployment
cs.RO cs.AI
Adaptive teaming, the ability to collaborate with unseen teammates without prior coordination, remains an underexplored challenge in multi-robot collaboration. This paper focuses on adaptive teaming in multi-drone cooperative pursuit, a critical task with real-world applications such as border surveillance, search-and-rescue, and counter-terrorism. We first define and formalize the \textbf{A}daptive Teaming in \textbf{M}ulti-\textbf{D}rone \textbf{P}ursuit (AT-MDP) problem and introduce AT-MDP framework, a comprehensive framework that integrates simulation, algorithm training and real-world deployment. AT-MDP framework provides a flexible experiment configurator and interface for simulation, a distributed training framework with an extensive algorithm zoo (including two newly proposed baseline methods) and an unseen drone zoo for evaluating adaptive teaming, as well as a real-world deployment system that utilizes edge computing and Crazyflie drones. To the best of our knowledge, AT-MDP framework is the first adaptive framework for continuous-action decision-making in complex real-world drone tasks, enabling multiple drones to coordinate effectively with unseen teammates. Extensive experiments in four multi-drone pursuit environments of increasing difficulty confirm the effectiveness of AT-MDP framework, while real-world deployments further validate its feasibility in physical systems. Videos and code are available at https://sites.google.com/view/at-mdp.
2502.09765
Differential Adjusted Parity for Learning Fair Representations
cs.LG cs.AI
The development of fair and unbiased machine learning models remains an ongoing objective for researchers in the field of artificial intelligence. We introduce the Differential Adjusted Parity (DAP) loss to produce unbiased informative representations. It utilises a differentiable variant of the adjusted parity metric to create a unified objective function. By combining downstream task classification accuracy and its inconsistency across sensitive feature domains, it provides a single tool to increase performance and mitigate bias. A key element in this approach is the use of soft balanced accuracies. In contrast to previous non-adversarial approaches, DAP does not suffer a degeneracy where the metric is satisfied by performing equally poorly across all sensitive domains. It outperforms several adversarial models on downstream task accuracy and fairness in our analysis. Specifically, it improves the demographic parity, equalized odds and sensitive feature accuracy by as much as 22.5\%, 44.1\% and 40.1\%, respectively, when compared to the best performing adversarial approaches on these metrics. Overall, the DAP loss and its associated metric can play a significant role in creating more fair machine learning models.
2502.09767
Non-Markovian Discrete Diffusion with Causal Language Models
cs.LG cs.AI cs.CL
Discrete diffusion models have emerged as a flexible and controllable paradigm for structured sequence modeling, yet they still lag behind causal language models in expressiveness. To bridge the gap between two paradigms, we introduce CaDDi, a causal discrete diffusion model that unifies sequential and temporal modeling within a non-Markovian diffusion framework. Unlike conventional diffusion models that operate step by step with no access to prior states, CaDDi integrates the temporal trajectory, enabling more expressive and controllable generation. Our approach also treats causal language models as a special case, allowing seamless adoption of pretrained large language models (LLMs) for discrete diffusion without the need for architectural modifications. Empirically, we demonstrate that CaDDi outperforms state-of-the-art discrete diffusion models on both natural language and biological sequence tasks, narrowing the gap between diffusion-based methods and large-scale autoregressive transformers.
2502.09768
Complex Network Modelling with Power-law Activating Patterns and Its Evolutionary Dynamics
cs.SI physics.soc-ph
Complex network theory provides a unifying framework for the study of structured dynamic systems. The current literature emphasizes a widely reported phenomenon of intermittent interaction among network vertices. In this paper, we introduce a complex network model that considers the stochastic switching of individuals between activated and quiescent states at power-law rates and the corresponding evolutionary dynamics. By using the Markov chain and renewal theory, we discover a homogeneous stationary distribution of activated sizes in the network with power-law activating patterns and infer some statistical characteristics. To better understand the effect of power-law activating patterns, we study the two-person-two-strategy evolutionary game dynamics, demonstrate the absorbability of strategies, and obtain the critical cooperation conditions for prisoner's dilemmas in homogeneous networks without mutation. The evolutionary dynamics in real networks are also discussed. Our results provide a new perspective to analyze and understand social physics in time-evolving network systems.
2502.09775
CellFlow: Simulating Cellular Morphology Changes via Flow Matching
q-bio.QM cs.CV cs.LG q-bio.BM q-bio.CB
Building a virtual cell capable of accurately simulating cellular behaviors in silico has long been a dream in computational biology. We introduce CellFlow, an image-generative model that simulates cellular morphology changes induced by chemical and genetic perturbations using flow matching. Unlike prior methods, CellFlow models distribution-wise transformations from unperturbed to perturbed cell states, effectively distinguishing actual perturbation effects from experimental artifacts such as batch effects -- a major challenge in biological data. Evaluated on chemical (BBBC021), genetic (RxRx1), and combined perturbation (JUMP) datasets, CellFlow generates biologically meaningful cell images that faithfully capture perturbation-specific morphological changes, achieving a 35% improvement in FID scores and a 12% increase in mode-of-action prediction accuracy over existing methods. Additionally, CellFlow enables continuous interpolation between cellular states, providing a potential tool for studying perturbation dynamics. These capabilities mark a significant step toward realizing virtual cell modeling for biomedical research.
2502.09777
On the existence of EFX allocations in multigraphs
cs.GT cs.AI
We study the problem of "fairly" dividing indivisible goods to several agents that have valuation set functions over the sets of goods. As fair we consider the allocations that are envy-free up to any good (EFX), i.e., no agent envies any proper subset of the goods given to any other agent. The existence or not of EFX allocations is a major open problem in Fair Division, and there are only positive results for special cases. [George Christodoulou, Amos Fiat, Elias Koutsoupias, Alkmini Sgouritsa 2023] introduced a restriction on the agents' valuations according to a graph structure: the vertices correspond to agents and the edges to goods, and each vertex/agent has zero marginal value (or in other words, they are indifferent) for the edges/goods that are not adjacent to them. The existence of EFX allocations has been shown for simple graphs with general monotone valuations [George Christodoulou, Amos Fiat, Elias Koutsoupias, Alkmini Sgouritsa 2023], and for multigraphs for restricted additive valuations [Alireza Kaviani, Masoud Seddighin, Amir Mohammad Shahrezaei 2024]. In this work, we push the state-of-the-art further, and show that the EFX allocations always exists in multigraphs and general monotone valuations if any of the following three conditions hold: either (a) the multigraph is bipartite, or (b) each agent has at most $\lceil \frac{n}{4} \rceil -1$ neighbors, where $n$ is the total number of agents, or (c) the shortest cycle with non-parallel edges has length at least 6.
2502.09778
Prompt and circumstance: A word-by-word LLM prompting approach to interlinear glossing for low-resource languages
cs.CL
Partly automated creation of interlinear glossed text (IGT) has the potential to assist in linguistic documentation. We argue that LLMs can make this process more accessible to linguists because of their capacity to follow natural-language instructions. We investigate the effectiveness of a retrieval-based LLM prompting approach to glossing, applied to the seven languages from the SIGMORPHON 2023 shared task. Our system beats the BERT-based shared task baseline for every language in the morpheme-level score category, and we show that a simple 3-best oracle has higher word-level scores than the challenge winner (a tuned sequence model) in five languages. In a case study on Tsez, we ask the LLM to automatically create and follow linguistic instructions, reducing errors on a confusing grammatical feature. Our results thus demonstrate the potential contributions which LLMs can make in interactive systems for glossing, both in making suggestions to human annotators and following directions.
2502.09779
Automated Muscle and Fat Segmentation in Computed Tomography for Comprehensive Body Composition Analysis
eess.IV cs.CV
Body composition assessment using CT images can potentially be used for a number of clinical applications, including the prognostication of cardiovascular outcomes, evaluation of metabolic health, monitoring of disease progression, assessment of nutritional status, prediction of treatment response in oncology, and risk stratification for surgical and critical care outcomes. While multiple groups have developed in-house segmentation tools for this analysis, there are very limited publicly available tools that could be consistently used across different applications. To mitigate this gap, we present a publicly accessible, end-to-end segmentation and feature calculation model specifically for CT body composition analysis. Our model performs segmentation of skeletal muscle, subcutaneous adipose tissue (SAT), and visceral adipose tissue (VAT) across the chest, abdomen, and pelvis area in axial CT images. It also provides various body composition metrics, including muscle density, visceral-to-subcutaneous fat (VAT/SAT) ratio, muscle area/volume, and skeletal muscle index (SMI), supporting both 2D and 3D assessments. The model is shared for public use. To evaluate the model, the segmentation was applied to both internal and external datasets, with body composition metrics analyzed across different age, sex, and race groups. The model achieved high dice coefficients on both internal and external datasets, exceeding 89% for skeletal muscle, SAT, and VAT segmentation. The model outperforms the benchmark by 2.40% on skeletal muscle and 10.26% on SAT compared to the manual annotations given by the publicly available dataset. Body composition metrics show mean relative absolute errors (MRAEs) under 10% for all measures. Furthermore, the model provided muscular fat segmentation with a Dice coefficient of 56.27%, which can be utilized for additional analyses as needed.
2502.09780
Incentivize without Bonus: Provably Efficient Model-based Online Multi-agent RL for Markov Games
cs.LG cs.AI cs.GT math.OC
Multi-agent reinforcement learning (MARL) lies at the heart of a plethora of applications involving the interaction of a group of agents in a shared unknown environment. A prominent framework for studying MARL is Markov games, with the goal of finding various notions of equilibria in a sample-efficient manner, such as the Nash equilibrium (NE) and the coarse correlated equilibrium (CCE). However, existing sample-efficient approaches either require tailored uncertainty estimation under function approximation, or careful coordination of the players. In this paper, we propose a novel model-based algorithm, called VMG, that incentivizes exploration via biasing the empirical estimate of the model parameters towards those with a higher collective best-response values of all the players when fixing the other players' policies, thus encouraging the policy to deviate from its current equilibrium for more exploration. VMG is oblivious to different forms of function approximation, and permits simultaneous and uncoupled policy updates of all players. Theoretically, we also establish that VMG achieves a near-optimal regret for finding both the NEs of two-player zero-sum Markov games and CCEs of multi-player general-sum Markov games under linear function approximation in an online environment, which nearly match their counterparts with sophisticated uncertainty quantification.
2502.09781
Medical Applications of Graph Convolutional Networks Using Electronic Health Records: A Survey
cs.LG
Graph Convolutional Networks (GCNs) have emerged as a promising approach to machine learning on Electronic Health Records (EHRs). By constructing a graph representation of patient data and performing convolutions on neighborhoods of nodes, GCNs can capture complex relationships and extract meaningful insights to support medical decision making. This survey provides an overview of the current research in applying GCNs to EHR data. We identify the key medical domains and prediction tasks where these models are being utilized, common benchmark datasets, and architectural patterns to provide a comprehensive survey of this field. While this is a nascent area of research, GCNs demonstrate strong potential to leverage the complex information hidden in EHRs. Challenges and opportunities for future work are also discussed.
2502.09782
Improving Acoustic Side-Channel Attacks on Keyboards Using Transformers and Large Language Models
cs.LG cs.AI cs.CL eess.AS
The increasing prevalence of microphones in everyday devices and the growing reliance on online services have amplified the risk of acoustic side-channel attacks (ASCAs) targeting keyboards. This study explores deep learning techniques, specifically vision transformers (VTs) and large language models (LLMs), to enhance the effectiveness and applicability of such attacks. We present substantial improvements over prior research, with the CoAtNet model achieving state-of-the-art performance. Our CoAtNet shows a 5.0% improvement for keystrokes recorded via smartphone (Phone) and 5.9% for those recorded via Zoom compared to previous benchmarks. We also evaluate transformer architectures and language models, with the best VT model matching CoAtNet's performance. A key advancement is the introduction of a noise mitigation method for real-world scenarios. By using LLMs for contextual understanding, we detect and correct erroneous keystrokes in noisy environments, enhancing ASCA performance. Additionally, fine-tuned lightweight language models with Low-Rank Adaptation (LoRA) deliver comparable performance to heavyweight models with 67X more parameters. This integration of VTs and LLMs improves the practical applicability of ASCA mitigation, marking the first use of these technologies to address ASCAs and error correction in real-world scenarios.
2502.09787
TableTalk: Scaffolding Spreadsheet Development with a Language Agent
cs.SE cs.AI cs.HC
Despite its ubiquity in the workforce, spreadsheet programming remains challenging as programmers need both spreadsheet-specific knowledge (e.g., APIs to write formulas) and problem-solving skills to create complex spreadsheets. Large language models (LLMs) can help automate aspects of this process, and recent advances in planning and reasoning have enabled language agents, which dynamically plan, use tools, and take iterative actions to complete complex tasks. These agents observe, plan, and act, making them well-suited to scaffold spreadsheet programming by following expert processes. We present TableTalk, a language agent that helps programmers build spreadsheets conversationally. Its design reifies three design principles -- scaffolding, flexibility, and incrementality -- which we derived from two studies of seven programmers and 62 Excel templates. TableTalk structures spreadsheet development by generating step-by-step plans and suggesting three next steps users can choose from. It also integrates tools that enable incremental spreadsheet construction. A user study with 20 programmers shows that TableTalk produces spreadsheets 2.3 times more likely to be preferred over a baseline agent, while reducing cognitive load and time spent reasoning about spreadsheet actions by 12.6%. TableTalk's approach has implications for human-agent collaboration. This includes providing persistent direct manipulation interfaces for stopping or undoing agent actions, while ensuring that such interfaces for accepting actions can be deactivated.
2502.09790
ExoMiner++ on TESS with Transfer Learning from Kepler: Transit Classification and Vetting Catalog for 2-min Data
astro-ph.EP astro-ph.IM cs.LG
We present ExoMiner++, an enhanced deep learning model that builds on the success of ExoMiner to improve transit signal classification in 2-minute TESS data. ExoMiner++ incorporates additional diagnostic inputs, including periodogram, flux trend, difference image, unfolded flux, and spacecraft attitude control data, all of which are crucial for effectively distinguishing transit signals from more challenging sources of false positives. To further enhance performance, we leverage transfer learning from high-quality labeled data from the Kepler space telescope, mitigating the impact of TESS's noisier and more ambiguous labels. ExoMiner++ achieves high accuracy across various classification and ranking metrics, significantly narrowing the search space for follow-up investigations to confirm new planets. To serve the exoplanet community, we introduce new TESS catalogs containing ExoMiner++ classifications and confidence scores for each transit signal. Among the 147,568 unlabeled TCEs, ExoMiner++ identifies 7,330 as planet candidates, with the remainder classified as false positives. These 7,330 planet candidates correspond to 1,868 existing TESS Objects of Interest (TOIs), 69 Community TESS Objects of Interest (CTOIs), and 50 newly introduced CTOIs. 1,797 out of the 2,506 TOIs previously labeled as planet candidates in ExoFOP are classified as planet candidates by ExoMiner++. This reduction in plausible candidates combined with the excellent ranking quality of ExoMiner++ allows the follow-up efforts to be focused on the most likely candidates, increasing the overall planet yield.
2502.09791
Atom identification in bilayer moire materials with Gomb-Net
cond-mat.mtrl-sci cs.CV
Moire patterns in van der Waals bilayer materials complicate the analysis of atomic-resolution images, hindering the atomic-scale insight typically attainable with scanning transmission electron microscopy. Here, we report a method to detect the positions and identity of atoms in each of the individual layers that compose bilayer heterostructures. We developed a deep learning model, Gomb-Net, which can distinguish atomic species in each individual layer, effectively deconvoluting the moire pattern to enable layer-specific mapping of strain and dopant distributions, unlike other methods which struggle with moire-induced complexity. Using this approach, we explored Se atom substitutional sites in a twisted fractional Janus WS2-WS2(1-x)Se2x heterostructure and found that layer specific implantation sites are unaffected by the moire pattern's local energetic or electronic modulation. This advancement enables atom-identification within material regimes where it was not possible before, opening new insights into previously inaccessible material physics.