id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2501.06003
Learning to generate feasible graphs using graph grammars
cs.LG
Generative methods for graphs need to be sufficiently flexible to model complex dependencies between sets of nodes. At the same time, the generated graphs need to satisfy domain-dependent feasibility conditions, that is, they should not violate certain constraints that would make their interpretation impossible within the given application domain (e.g. a molecular graph where an atom has a very large number of chemical bounds). Crucially, constraints can involve not only local but also long-range dependencies: for example, the maximal length of a cycle can be bounded. Currently, a large class of generative approaches for graphs, such as methods based on artificial neural networks, is based on message passing schemes. These approaches suffer from information 'dilution' issues that severely limit the maximal range of the dependencies that can be modeled. To address this problem, we propose a generative approach based on the notion of graph grammars. The key novel idea is to introduce a domain-dependent coarsening procedure to provide short-cuts for long-range dependencies. We show the effectiveness of our proposal in two domains: 1) small drugs and 2) RNA secondary structures. In the first case, we compare the quality of the generated molecular graphs via the Molecular Sets (MOSES) benchmark suite, which evaluates the distance between generated and real molecules, their lipophilicity, synthesizability, and drug-likeness. In the second case, we show that the approach can generate very large graphs (with hundreds of nodes) that are accepted as valid examples for a desired RNA family by the "Infernal" covariance model, a state-of-the-art RNA classifier. Our implementation is available on github: github.com/fabriziocosta/GraphLearn
2501.06004
SeMi: When Imbalanced Semi-Supervised Learning Meets Mining Hard Examples
cs.CV
Semi-Supervised Learning (SSL) can leverage abundant unlabeled data to boost model performance. However, the class-imbalanced data distribution in real-world scenarios poses great challenges to SSL, resulting in performance degradation. Existing class-imbalanced semi-supervised learning (CISSL) methods mainly focus on rebalancing datasets but ignore the potential of using hard examples to enhance performance, making it difficult to fully harness the power of unlabeled data even with sophisticated algorithms. To address this issue, we propose a method that enhances the performance of Imbalanced Semi-Supervised Learning by Mining Hard Examples (SeMi). This method distinguishes the entropy differences among logits of hard and easy examples, thereby identifying hard examples and increasing the utility of unlabeled data, better addressing the imbalance problem in CISSL. In addition, we maintain a class-balanced memory bank with confidence decay for storing high-confidence embeddings to enhance the pseudo-labels' reliability. Although our method is simple, it is effective and seamlessly integrates with existing approaches. We perform comprehensive experiments on standard CISSL benchmarks and experimentally demonstrate that our proposed SeMi outperforms existing state-of-the-art methods on multiple benchmarks, especially in reversed scenarios, where our best result shows approximately a 54.8\% improvement over the baseline methods.
2501.06006
CamCtrl3D: Single-Image Scene Exploration with Precise 3D Camera Control
cs.CV
We propose a method for generating fly-through videos of a scene, from a single image and a given camera trajectory. We build upon an image-to-video latent diffusion model. We condition its UNet denoiser on the camera trajectory, using four techniques. (1) We condition the UNet's temporal blocks on raw camera extrinsics, similar to MotionCtrl. (2) We use images containing camera rays and directions, similar to CameraCtrl. (3) We reproject the initial image to subsequent frames and use the resulting video as a condition. (4) We use 2D<=>3D transformers to introduce a global 3D representation, which implicitly conditions on the camera poses. We combine all conditions in a ContolNet-style architecture. We then propose a metric that evaluates overall video quality and the ability to preserve details with view changes, which we use to analyze the trade-offs of individual and combined conditions. Finally, we identify an optimal combination of conditions. We calibrate camera positions in our datasets for scale consistency across scenes, and we train our scene exploration model, CamCtrl3D, demonstrating state-of-theart results.
2501.06007
CoNOAir: A Neural Operator for Forecasting Carbon Monoxide Evolution in Cities
cs.LG
Carbon Monoxide (CO) is a dominant pollutant in urban areas due to the energy generation from fossil fuels for industry, automobile, and domestic requirements. Forecasting the evolution of CO in real-time can enable the deployment of effective early warning systems and intervention strategies. However, the computational cost associated with the physics and chemistry-based simulation makes it prohibitive to implement such a model at the city and country scale. To address this challenge, here, we present a machine learning model based on neural operator, namely, Complex Neural Operator for Air Quality (CoNOAir), that can effectively forecast CO concentrations. We demonstrate this by developing a country-level model for short-term (hourly) and long-term (72-hour) forecasts of CO concentrations. Our model outperforms state-of-the-art models such as Fourier neural operators (FNO) and provides reliable predictions for both short and long-term forecasts. We further analyse the capability of the model to capture extreme events and generate forecasts in urban cities in India. Interestingly, we observe that the model predicts the next hour CO concentrations with R2 values greater than 0.95 for all the cities considered. The deployment of such a model can greatly assist the governing bodies to provide early warning, plan intervention strategies, and develop effective strategies by considering several what-if scenarios. Altogether, the present approach could provide a fillip to real-time predictions of CO pollution in urban cities.
2501.06014
Pose-independent 3D Anthropometry from Sparse Data
cs.CV
3D digital anthropometry is the study of estimating human body measurements from 3D scans. Precise body measurements are important health indicators in the medical industry, and guiding factors in the fashion, ergonomic and entertainment industries. The measuring protocol consists of scanning the whole subject in the static A-pose, which is maintained without breathing or movement during the scanning process. However, the A-pose is not easy to maintain during the whole scanning process, which can last even up to a couple of minutes. This constraint affects the final quality of the scan, which in turn affects the accuracy of the estimated body measurements obtained from methods that rely on dense geometric data. Additionally, this constraint makes it impossible to develop a digital anthropometry method for subjects unable to assume the A-pose, such as those with injuries or disabilities. We propose a method that can obtain body measurements from sparse landmarks acquired in any pose. We make use of the sparse landmarks of the posed subject to create pose-independent features, and train a network to predict the body measurements as taken from the standard A-pose. We show that our method achieves comparable results to competing methods that use dense geometry in the standard A-pose, but has the capability of estimating the body measurements from any pose using sparse landmarks only. Finally, we address the lack of open-source 3D anthropometry methods by making our method available to the research community at https://github.com/DavidBoja/pose-independent-anthropometry.
2501.06016
Investigating the Impact of Observation Space Design Choices On Training Reinforcement Learning Solutions for Spacecraft Problems
cs.LG cs.SY eess.SY
Recent research using Reinforcement Learning (RL) to learn autonomous control for spacecraft operations has shown great success. However, a recent study showed their performance could be improved by changing the action space, i.e. control outputs, used in the learning environment. This has opened the door for finding more improvements through further changes to the environment. The work in this paper focuses on how changes to the environment's observation space can impact the training and performance of RL agents learning the spacecraft inspection task. The studies are split into two groups. The first looks at the impact of sensors that were designed to help agents learn the task. The second looks at the impact of reference frames, reorienting the agent to see the world from a different perspective. The results show the sensors are not necessary, but most of them help agents learn more optimal behavior, and that the reference frame does not have a large impact, but is best kept consistent.
2501.06019
BRIGHT: A globally distributed multimodal building damage assessment dataset with very-high-resolution for all-weather disaster response
cs.CV cs.AI eess.IV eess.SP
Disaster events occur around the world and cause significant damage to human life and property. Earth observation (EO) data enables rapid and comprehensive building damage assessment (BDA), an essential capability in the aftermath of a disaster to reduce human casualties and to inform disaster relief efforts. Recent research focuses on the development of AI models to achieve accurate mapping of unseen disaster events, mostly using optical EO data. However, solutions based on optical data are limited to clear skies and daylight hours, preventing a prompt response to disasters. Integrating multimodal (MM) EO data, particularly the combination of optical and SAR imagery, makes it possible to provide all-weather, day-and-night disaster responses. Despite this potential, the development of robust multimodal AI models has been constrained by the lack of suitable benchmark datasets. In this paper, we present a BDA dataset using veRy-hIGH-resoluTion optical and SAR imagery (BRIGHT) to support AI-based all-weather disaster response. To the best of our knowledge, BRIGHT is the first open-access, globally distributed, event-diverse MM dataset specifically curated to support AI-based disaster response. It covers five types of natural disasters and two types of man-made disasters across 12 regions worldwide, with a particular focus on developing countries where external assistance is most needed. The optical and SAR imagery in BRIGHT, with a spatial resolution between 0.3-1 meters, provides detailed representations of individual buildings, making it ideal for precise BDA. In our experiments, we have tested seven advanced AI models trained with our BRIGHT to validate the transferability and robustness. The dataset and code are available at https://github.com/ChenHongruixuan/BRIGHT. BRIGHT also serves as the official dataset for the 2025 IEEE GRSS Data Fusion Contest.
2501.06025
How to Tune a Multilingual Encoder Model for Germanic Languages: A Study of PEFT, Full Fine-Tuning, and Language Adapters
cs.CL cs.AI
This paper investigates the optimal use of the multilingual encoder model mDeBERTa for tasks in three Germanic languages -- German, Swedish, and Icelandic -- representing varying levels of presence and likely data quality in mDeBERTas pre-training data. We compare full fine-tuning with the parameter-efficient fine-tuning (PEFT) methods LoRA and Pfeiffer bottleneck adapters, finding that PEFT is more effective for the higher-resource language, German. However, results for Swedish and Icelandic are less consistent. We also observe differences between tasks: While PEFT tends to work better for question answering, full fine-tuning is preferable for named entity recognition. Inspired by previous research on modular approaches that combine task and language adapters, we evaluate the impact of adding PEFT modules trained on unstructured text, finding that this approach is not beneficial.
2501.06027
Geometric-Based Nail Segmentation for Clinical Measurements
cs.CV
A robust segmentation method that can be used to perform measurements on toenails is presented. The proposed method is used as the first step in a clinical trial to objectively quantify the incidence of a particular pathology. For such an assessment, it is necessary to distinguish a nail, which locally appears to be similar to the skin. Many algorithms have been used, each of which leverages different aspects of toenail appearance. We used the Hough transform to locate the tip of the toe and estimate the nail location and size. Subsequently, we classified the super-pixels of the image based on their geometric and photometric information. Thereafter, the watershed transform delineated the border of the nail. The method was validated using a 348-image medical dataset, achieving an accuracy of 0.993 and an F-measure of 0.925. The proposed method is considerably robust across samples, with respect to factors such as nail shape, skin pigmentation, illumination conditions, and appearance of large regions affected by a medical condition
2501.06030
Resiliency metrics quantifying emergency response in a distribution system
eess.SY cs.SY
The electric distribution system is a cornerstone of modern life, playing a critical role in the daily activities and well-being of individuals. As the world transitions toward a decarbonized future, where even mobility relies on electricity, ensuring the resilience of the grid becomes paramount. This paper introduces novel resilience metrics designed to equip utilities and stakeholders with actionable tools to assess performance during storm events. The metrics focus on emergency storm response and the resources required to improve customer service. The practical calculation of the metrics from historical utility data is demonstrated for multiple storm events. Additionally, the metrics' improvement with added crews is estimated by "rerunning history" with faster restoration. By applying this resilience framework, utilities can enhance their restoration strategies and unlock potential cost savings, benefiting both providers and customers in an era of heightened energy dependency.
2501.06031
Generate, Transduct, Adapt: Iterative Transduction with VLMs
cs.CV
Transductive zero-shot learning with vision-language models leverages image-image similarities within the dataset to achieve better classification accuracy compared to the inductive setting. However, there is little work that explores the structure of the language space in this context. We propose GTA-CLIP, a novel technique that incorporates supervision from language models for joint transduction in language and vision spaces. Our approach is iterative and consists of three steps: (i) incrementally exploring the attribute space by querying language models, (ii) an attribute-augmented transductive inference procedure, and (iii) fine-tuning the language and vision encoders based on inferred labels within the dataset. Through experiments with CLIP encoders, we demonstrate that GTA-CLIP, yields an average performance improvement of 8.6% and 3.7% across 12 datasets and 3 encoders, over CLIP and transductive CLIP respectively in the zero-shot setting. We also observe similar improvements in a few-shot setting. We present ablation studies that demonstrate the value of each step and visualize how the vision and language spaces evolve over iterations driven by the transductive learning.
2501.06035
Nonisotropic Gaussian Diffusion for Realistic 3D Human Motion Prediction
cs.CV
Probabilistic human motion prediction aims to forecast multiple possible future movements from past observations. While current approaches report high diversity and realism, they often generate motions with undetected limb stretching and jitter. To address this, we introduce SkeletonDiffusion, a latent diffusion model that embeds an explicit inductive bias on the human body within its architecture and training. Our model is trained with a novel nonisotropic Gaussian diffusion formulation that aligns with the natural kinematic structure of the human skeleton. Results show that our approach outperforms conventional isotropic alternatives, consistently generating realistic predictions while avoiding artifacts such as limb distortion. Additionally, we identify a limitation in commonly used diversity metrics, which may inadvertently favor models that produce inconsistent limb lengths within the same sequence. SkeletonDiffusion sets a new benchmark on three real-world datasets, outperforming various baselines across multiple evaluation metrics. Visit our project page: https://ceveloper.github.io/publications/skeletondiffusion/
2501.06038
A Holistically Point-guided Text Framework for Weakly-Supervised Camouflaged Object Detection
cs.CV
Weakly-Supervised Camouflaged Object Detection (WSCOD) has gained popularity for its promise to train models with weak labels to segment objects that visually blend into their surroundings. Recently, some methods using sparsely-annotated supervision shown promising results through scribbling in WSCOD, while point-text supervision remains underexplored. Hence, this paper introduces a novel holistically point-guided text framework for WSCOD by decomposing into three phases: segment, choose, train. Specifically, we propose Point-guided Candidate Generation (PCG), where the point's foreground serves as a correction for the text path to explicitly correct and rejuvenate the loss detection object during the mask generation process (SEGMENT). We also introduce a Qualified Candidate Discriminator (QCD) to choose the optimal mask from a given text prompt using CLIP (CHOOSE), and employ the chosen pseudo mask for training with a self-supervised Vision Transformer (TRAIN). Additionally, we developed a new point-supervised dataset (P2C-COD) and a text-supervised dataset (T-COD). Comprehensive experiments on four benchmark datasets demonstrate our method outperforms state-of-the-art methods by a large margin, and also outperforms some existing fully-supervised camouflaged object detection methods.
2501.06039
AI-powered virtual tissues from spatial proteomics for clinical diagnostics and biomedical discovery
q-bio.QM cs.AI cs.CV cs.LG
Spatial proteomics technologies have transformed our understanding of complex tissue architectures by enabling simultaneous analysis of multiple molecular markers and their spatial organization. The high dimensionality of these data, varying marker combinations across experiments and heterogeneous study designs pose unique challenges for computational analysis. Here, we present Virtual Tissues (VirTues), a foundation model framework for biological tissues that operates across the molecular, cellular and tissue scale. VirTues introduces innovations in transformer architecture design, including a novel tokenization scheme that captures both spatial and marker dimensions, and attention mechanisms that scale to high-dimensional multiplex data while maintaining interpretability. Trained on diverse cancer and non-cancer tissue datasets, VirTues demonstrates strong generalization capabilities without task-specific fine-tuning, enabling cross-study analysis and novel marker integration. As a generalist model, VirTues outperforms existing approaches across clinical diagnostics, biological discovery and patient case retrieval tasks, while providing insights into tissue function and disease mechanisms.
2501.06040
MSCViT: A Small-size ViT architecture with Multi-Scale Self-Attention Mechanism for Tiny Datasets
cs.CV
Vision Transformer (ViT) has demonstrated significant potential in various vision tasks due to its strong ability in modelling long-range dependencies. However, such success is largely fueled by training on massive samples. In real applications, the large-scale datasets are not always available, and ViT performs worse than Convolutional Neural Networks (CNNs) if it is only trained on small scale dataset (called tiny dataset), since it requires large amount of training data to ensure its representational capacity. In this paper, a small-size ViT architecture with multi-scale self-attention mechanism and convolution blocks is presented (dubbed MSCViT) to model different scales of attention at each layer. Firstly, we introduced wavelet convolution, which selectively combines the high-frequency components obtained by frequency division with our convolution channel to extract local features. Then, a lightweight multi-head attention module is developed to reduce the number of tokens and computational costs. Finally, the positional encoding (PE) in the backbone is replaced by a local feature extraction module. Compared with the original ViT, it is parameter-efficient and is particularly suitable for tiny datasets. Extensive experiments have been conducted on tiny datasets, in which our model achieves an accuracy of 84.68% on CIFAR-100 with 14.0M parameters and 2.5 GFLOPs, without pre-training on large datasets.
2501.06042
The improvement in transmission resilience metrics from reduced outages or faster restoration can be calculated by rerunning historical outage data
physics.soc-ph cs.SY eess.SY
Transmission utilities routinely collect detailed outage data, including resilience events in which outages bunch up due to weather. The resilience events and their resilience metrics can readily be extracted from this historical outage data. Improvements such as grid hardening or investments in restoration lead to reduced outages or faster restoration. We show how to rerun this history with the effects of the reduced outages or faster restoration included to find the resulting improvement in resilience metrics, thus quantifying the benefits of these investments. This is demonstrated with case studies for specific events (a derecho and a hurricane), and all large events or large thunderstorms in the Midwest USA. Instead of predicting future extreme events with models, which is very challenging, the historical rerun readily quantifies the benefits that a resilience investment would have had if it had been made in the past. The historical rerun is particularly vivid in making the case for resilience investments to stakeholders because it quantifies the benefits for events actually experienced by those stakeholders, rather than for future events predicted with uncertainty.
2501.06047
Learning Affordances from Interactive Exploration using an Object-level Map
cs.RO
Many robotic tasks in real-world environments require physical interactions with an object such as pick up or push. For successful interactions, the robot needs to know the object's affordances, which are defined as the potential actions the robot can perform with the object. In order to learn a robot-specific affordance predictor, we propose an interactive exploration pipeline which allows the robot to collect interaction experiences while exploring an unknown environment. We integrate an object-level map in the exploration pipeline such that the robot can identify different object instances and track objects across diverse viewpoints. This results in denser and more accurate affordance annotations compared to state-of-the-art methods, which do not incorporate a map. We show that our affordance exploration approach makes exploration more efficient and results in more accurate affordance prediction models compared to baseline methods.
2501.06051
Benchmarking Rotary Position Embeddings for Automatic Speech Recognition
cs.CL cs.AI eess.AS
Rotary Position Embedding (RoPE) encodes relative and absolute positional information in Transformer-based models through rotation matrices applied to input vectors within sequences. While RoPE has demonstrated superior performance compared to other positional embedding technologies in natural language processing tasks, its effectiveness in speech processing applications remains understudied. In this work, we conduct a comprehensive evaluation of RoPE across diverse automatic speech recognition (ASR) tasks. Our experimental results demonstrate that for ASR tasks, RoPE consistently achieves lower error rates compared to the currently widely used relative positional embedding. To facilitate further research, we release the implementation and all experimental recipes through the SpeechBrain toolkit.
2501.06053
Enhancing, Refining, and Fusing: Towards Robust Multi-Scale and Dense Ship Detection
cs.CV
Synthetic aperture radar (SAR) imaging, celebrated for its high resolution, all-weather capability, and day-night operability, is indispensable for maritime applications. However, ship detection in SAR imagery faces significant challenges, including complex backgrounds, densely arranged targets, and large scale variations. To address these issues, we propose a novel framework, Center-Aware SAR Ship Detector (CASS-Det), designed for robust multi-scale and densely packed ship detection. CASS-Det integrates three key innovations: (1) a center enhancement module (CEM) that employs rotational convolution to emphasize ship centers, improving localization while suppressing background interference; (2) a neighbor attention module (NAM) that leverages cross-layer dependencies to refine ship boundaries in densely populated scenes; and (3) a cross-connected feature pyramid network (CC-FPN) that enhances multi-scale feature fusion by integrating shallow and deep features. Extensive experiments on the SSDD, HRSID, and LS-SSDD-v1.0 datasets demonstrate the state-of-the-art performance of CASS-Det, excelling at detecting multi-scale and densely arranged ships.
2501.06058
Learning Flexible Heterogeneous Coordination with Capability-Aware Shared Hypernetworks
cs.MA cs.LG
Cooperative heterogeneous multi-agent tasks require agents to effectively coordinate their behaviors while accounting for their relative capabilities. Learning-based solutions to this challenge span between two extremes: i) shared-parameter methods, which encode diverse behaviors within a single architecture by assigning an ID to each agent, and are sample-efficient but result in limited behavioral diversity; ii) independent methods, which learn a separate policy for each agent, and show greater behavioral diversity but lack sample-efficiency. Prior work has also explored selective parameter-sharing, allowing for a compromise between diversity and efficiency. None of these approaches, however, effectively generalize to unseen agents or teams. We present Capability-Aware Shared Hypernetworks (CASH), a novel architecture for heterogeneous multi-agent coordination that generates sufficient diversity while maintaining sample-efficiency via soft parameter-sharing hypernetworks. Intuitively, CASH allows the team to learn common strategies using a shared encoder, which are then adapted according to the team's individual and collective capabilities with a hypernetwork, allowing for zero-shot generalization to unseen teams and agents. We present experiments across two heterogeneous coordination tasks and three standard learning paradigms (imitation learning, on- and off-policy reinforcement learning). CASH is able to outperform baseline architectures in success rate and sample efficiency when evaluated on unseen teams and agents despite using less than half of the learnable parameters.
2501.06059
COMIX: Compositional Explanations using Prototypes
cs.LG
Aligning machine representations with human understanding is key to improving interpretability of machine learning (ML) models. When classifying a new image, humans often explain their decisions by decomposing the image into concepts and pointing to corresponding regions in familiar images. Current ML explanation techniques typically either trace decision-making processes to reference prototypes, generate attribution maps highlighting feature importance, or incorporate intermediate bottlenecks designed to align with human-interpretable concepts. The proposed method, named COMIX, classifies an image by decomposing it into regions based on learned concepts and tracing each region to corresponding ones in images from the training dataset, assuring that explanations fully represent the actual decision-making process. We dissect the test image into selected internal representations of a neural network to derive prototypical parts (primitives) and match them with the corresponding primitives derived from the training data. In a series of qualitative and quantitative experiments, we theoretically prove and demonstrate that our method, in contrast to post hoc analysis, provides fidelity of explanations and shows that the efficiency is competitive with other inherently interpretable architectures. Notably, it shows substantial improvements in fidelity and sparsity metrics, including 48.82% improvement in the C-insertion score on the ImageNet dataset over the best state-of-the-art baseline.
2501.06062
Personalized Language Model Learning on Text Data Without User Identifiers
cs.LG
In many practical natural language applications, user data are highly sensitive, requiring anonymous uploads of text data from mobile devices to the cloud without user identifiers. However, the absence of user identifiers restricts the ability of cloud-based language models to provide personalized services, which are essential for catering to diverse user needs. The trivial method of replacing an explicit user identifier with a static user embedding as model input still compromises data anonymization. In this work, we propose to let each mobile device maintain a user-specific distribution to dynamically generate user embeddings, thereby breaking the one-to-one mapping between an embedding and a specific user. We further theoretically demonstrate that to prevent the cloud from tracking users via uploaded embeddings, the local distributions of different users should either be derived from a linearly dependent space to avoid identifiability or be close to each other to prevent accurate attribution. Evaluation on both public and industrial datasets using different language models reveals a remarkable improvement in accuracy from incorporating anonymous user embeddings, while preserving real-time inference requirement.
2501.06066
Distilling Calibration via Conformalized Credal Inference
cs.LG cs.AI eess.SP
Deploying artificial intelligence (AI) models on edge devices involves a delicate balance between meeting stringent complexity constraints, such as limited memory and energy resources, and ensuring reliable performance in sensitive decision-making tasks. One way to enhance reliability is through uncertainty quantification via Bayesian inference. This approach, however, typically necessitates maintaining and running multiple models in an ensemble, which may exceed the computational limits of edge devices. This paper introduces a low-complexity methodology to address this challenge by distilling calibration information from a more complex model. In an offline phase, predictive probabilities generated by a high-complexity cloud-based model are leveraged to determine a threshold based on the typical divergence between the cloud and edge models. At run time, this threshold is used to construct credal sets -- ranges of predictive probabilities that are guaranteed, with a user-selected confidence level, to include the predictions of the cloud model. The credal sets are obtained through thresholding of a divergence measure in the simplex of predictive probabilities. Experiments on visual and language tasks demonstrate that the proposed approach, termed Conformalized Distillation for Credal Inference (CD-CI), significantly improves calibration performance compared to low-complexity Bayesian methods, such as Laplace approximation, making it a practical and efficient solution for edge AI deployments.
2501.06067
Decentralized Multi-Antenna Architectures with Unitary Constraints
eess.SP cs.IT math.IT
The increase in the number of base station (BS) antennas calls for efficient solutions to deal with the increased interconnection bandwidth and processing complexity of traditional centralized approaches. Decentralized approaches are thus gaining momentum, since they achieve important reductions in data/processing volume by preprocessing the received signals before forwarding them to a central node. The WAX framework offers a general description of decentralized architectures with arbitrary interplay between interconnection bandwidth and decentralized processing complexity, but the applicability of this framework has only been studied assuming unrestricted baseband processing. We consider an adaptation of the WAX framework where the decentralized processing has unitary restriction, which allows for energy-efficient implementations based on reconfigurable impedance networks at the cost of some performance loss. Moreover, we propose an effective method to minimize the performance gap with respect to centralized processing. The previous method gives a first step towards characterizing the information-lossless trade-off between interconnection bandwidth and processing complexity in decentralized architectures with unitary constraints.
2501.06074
Geometry and Optimization of Shallow Polynomial Networks
cs.LG math.AG
We study shallow neural networks with polynomial activations. The function space for these models can be identified with a set of symmetric tensors with bounded rank. We describe general features of these networks, focusing on the relationship between width and optimization. We then consider teacher-student problems, that can be viewed as a problem of low-rank tensor approximation with respect to a non-standard inner product that is induced by the data distribution. In this setting, we introduce a teacher-metric discriminant which encodes the qualitative behavior of the optimization as a function of the training data distribution. Finally, we focus on networks with quadratic activations, presenting an in-depth analysis of the optimization landscape. In particular, we present a variation of the Eckart-Young Theorem characterizing all critical points and their Hessian signatures for teacher-student problems with quadratic networks and Gaussian training data.
2501.06076
A monthly sub-national Harmonized Food Insecurity Dataset for comprehensive analysis and predictive modeling
cs.LG
Food security is a complex, multidimensional concept challenging to measure comprehensively. Effective anticipation, monitoring, and mitigation of food crises require timely and comprehensive global data. This paper introduces the Harmonized Food Insecurity Dataset (HFID), an open-source resource consolidating four key data sources: the Integrated Food Security Phase Classification (IPC)/Cadre Harmonis\'e (CH) phases, the Famine Early Warning Systems Network (FEWS NET) IPC-compatible phases, and the World Food Program's (WFP) Food Consumption Score (FCS) and reduced Coping Strategy Index (rCSI). Updated monthly and using a common reference system for administrative units, the HFID offers extensive spatial and temporal coverage. It serves as a vital tool for food security experts and humanitarian agencies, providing a unified resource for analyzing food security conditions and highlighting global data disparities. The scientific community can also leverage the HFID to develop data-driven predictive models, enhancing the capacity to forecast and prevent future food crises.
2501.06077
Explainable Federated Bayesian Causal Inference and Its Application in Advanced Manufacturing
cs.LG stat.AP
Causal inference has recently gained notable attention across various fields like biology, healthcare, and environmental science, especially within explainable artificial intelligence (xAI) systems, for uncovering the causal relationships among multiple variables and outcomes. Yet, it has not been fully recognized and deployed in the manufacturing systems. In this paper, we introduce an explainable, scalable, and flexible federated Bayesian learning framework, \texttt{xFBCI}, designed to explore causality through treatment effect estimation in distributed manufacturing systems. By leveraging federated Bayesian learning, we efficiently estimate posterior of local parameters to derive the propensity score for each client without accessing local private data. These scores are then used to estimate the treatment effect using propensity score matching (PSM). Through simulations on various datasets and a real-world Electrohydrodynamic (EHD) printing data, we demonstrate that our approach outperforms standard Bayesian causal inference methods and several state-of-the-art federated learning benchmarks.
2501.06078
Explaining k-Nearest Neighbors: Abductive and Counterfactual Explanations
cs.LG cs.AI
Despite the wide use of $k$-Nearest Neighbors as classification models, their explainability properties remain poorly understood from a theoretical perspective. While nearest neighbors classifiers offer interpretability from a "data perspective", in which the classification of an input vector $\bar{x}$ is explained by identifying the vectors $\bar{v}_1, \ldots, \bar{v}_k$ in the training set that determine the classification of $\bar{x}$, we argue that such explanations can be impractical in high-dimensional applications, where each vector has hundreds or thousands of features and it is not clear what their relative importance is. Hence, we focus on understanding nearest neighbor classifications through a "feature perspective", in which the goal is to identify how the values of the features in $\bar{x}$ affect its classification. Concretely, we study abductive explanations such as "minimum sufficient reasons", which correspond to sets of features in $\bar{x}$ that are enough to guarantee its classification, and "counterfactual explanations" based on the minimum distance feature changes one would have to perform in $\bar{x}$ to change its classification. We present a detailed landscape of positive and negative complexity results for counterfactual and abductive explanations, distinguishing between discrete and continuous feature spaces, and considering the impact of the choice of distance function involved. Finally, we show that despite some negative complexity results, Integer Quadratic Programming and SAT solving allow for computing explanations in practice.
2501.06080
Scale-up Unlearnable Examples Learning with High-Performance Computing
cs.LG cs.AI cs.DC
Recent advancements in AI models are structured to retain user interactions, which could inadvertently include sensitive healthcare data. In the healthcare field, particularly when radiologists use AI-driven diagnostic tools hosted on online platforms, there is a risk that medical imaging data may be repurposed for future AI training without explicit consent, spotlighting critical privacy and intellectual property concerns around healthcare data usage. Addressing these privacy challenges, a novel approach known as Unlearnable Examples (UEs) has been introduced, aiming to make data unlearnable to deep learning models. A prominent method within this area, called Unlearnable Clustering (UC), has shown improved UE performance with larger batch sizes but was previously limited by computational resources. To push the boundaries of UE performance with theoretically unlimited resources, we scaled up UC learning across various datasets using Distributed Data Parallel (DDP) training on the Summit supercomputer. Our goal was to examine UE efficacy at high-performance computing (HPC) levels to prevent unauthorized learning and enhance data security, particularly exploring the impact of batch size on UE's unlearnability. Utilizing the robust computational capabilities of the Summit, extensive experiments were conducted on diverse datasets such as Pets, MedMNist, Flowers, and Flowers102. Our findings reveal that both overly large and overly small batch sizes can lead to performance instability and affect accuracy. However, the relationship between batch size and unlearnability varied across datasets, highlighting the necessity for tailored batch size strategies to achieve optimal data protection. Our results underscore the critical role of selecting appropriate batch sizes based on the specific characteristics of each dataset to prevent learning and ensure data security in deep learning applications.
2501.06081
Averaged Adam accelerates stochastic optimization in the training of deep neural network approximations for partial differential equation and optimal control problems
math.OC cs.LG cs.NA math.NA
Deep learning methods - usually consisting of a class of deep neural networks (DNNs) trained by a stochastic gradient descent (SGD) optimization method - are nowadays omnipresent in data-driven learning problems as well as in scientific computing tasks such as optimal control (OC) and partial differential equation (PDE) problems. In practically relevant learning tasks, often not the plain-vanilla standard SGD optimization method is employed to train the considered class of DNNs but instead more sophisticated adaptive and accelerated variants of the standard SGD method such as the popular Adam optimizer are used. Inspired by the classical Polyak-Ruppert averaging approach, in this work we apply averaged variants of the Adam optimizer to train DNNs to approximately solve exemplary scientific computing problems in the form of PDEs and OC problems. We test the averaged variants of Adam in a series of learning problems including physics-informed neural network (PINN), deep backward stochastic differential equation (deep BSDE), and deep Kolmogorov approximations for PDEs (such as heat, Black-Scholes, Burgers, and Allen-Cahn PDEs), including DNN approximations for OC problems, and including DNN approximations for image classification problems (ResNet for CIFAR-10). In each of the numerical examples the employed averaged variants of Adam outperform the standard Adam and the standard SGD optimizers, particularly, in the situation of the scientific machine learning problems. The Python source codes for the numerical experiments associated to this work can be found on GitHub at https://github.com/deeplearningmethods/averaged-adam.
2501.06086
All AI Models are Wrong, but Some are Optimal
cs.AI cs.LG
AI models that predict the future behavior of a system (a.k.a. predictive AI models) are central to intelligent decision-making. However, decision-making using predictive AI models often results in suboptimal performance. This is primarily because AI models are typically constructed to best fit the data, and hence to predict the most likely future rather than to enable high-performance decision-making. The hope that such prediction enables high-performance decisions is neither guaranteed in theory nor established in practice. In fact, there is increasing empirical evidence that predictive models must be tailored to decision-making objectives for performance. In this paper, we establish formal (necessary and sufficient) conditions that a predictive model (AI-based or not) must satisfy for a decision-making policy established using that model to be optimal. We then discuss their implications for building predictive AI models for sequential decision-making.
2501.06088
Non-planar 3D Printing of Double Shells
cs.RO cs.CG
We present a method to fabricate double shell structures printed in trans-versal directions using multi-axis fused-deposition-modeling (FDM) robot-ic 3D printing. Shell structures, characterized by lightweight, thin walls, fast buildup, and minimal material usage, find diverse applications in pro-totyping and architecture for uses such as fa\c{c}ade panels, molds for concrete casting, or full-scale pavilions. We leverage an underlying representation of transversal strip networks generated using existing methods and propose a methodology for converting them into printable partitions. Each partition is printed separately and assembled into a double-shell structure. We out-line the specifications and workflow that make the printing of each piece and the subsequent assembly process feasible. The versatility and robust-ness of our method are demonstrated with both digital and fabricated re-sults on surfaces of different scales and geometric complexity.
2501.06089
Towards Developing Socially Compliant Automated Vehicles: State of the Art, Experts Expectations, and A Conceptual Framework
cs.RO cs.AI cs.LG cs.MA cs.SY eess.SY
Automated Vehicles (AVs) hold promise for revolutionizing transportation by improving road safety, traffic efficiency, and overall mobility. Despite the steady advancement in high-level AVs in recent years, the transition to full automation entails a period of mixed traffic, where AVs of varying automation levels coexist with human-driven vehicles (HDVs). Making AVs socially compliant and understood by human drivers is expected to improve the safety and efficiency of mixed traffic. Thus, ensuring AVs compatibility with HDVs and social acceptance is crucial for their successful and seamless integration into mixed traffic. However, research in this critical area of developing Socially Compliant AVs (SCAVs) remains sparse. This study carries out the first comprehensive scoping review to assess the current state of the art in developing SCAVs, identifying key concepts, methodological approaches, and research gaps. An expert interview was also conducted to identify critical research gaps and expectations towards SCAVs. Based on the scoping review and expert interview input, a conceptual framework is proposed for the development of SCAVs. The conceptual framework is evaluated using an online survey targeting researchers, technicians, policymakers, and other relevant professionals worldwide. The survey results provide valuable validation and insights, affirming the significance of the proposed conceptual framework in tackling the challenges of integrating AVs into mixed-traffic environments. Additionally, future research perspectives and suggestions are discussed, contributing to the research and development agenda of SCAVs.
2501.06092
Molecular Communication-Inspired Particle Collector-Transmitter (PaCoT) for Heavy Metal Removal from Human Circulatory System
eess.SY cs.SY
This study proposes a novel molecular communication (MC)-inspired nanomachine, PArticle COllector-Transmitter (PaCoT), to remove toxic heavy metals from the human circulatory system. PaCoT collects these toxic metals and transmits them to release nodes, such as lymph capillaries, before they reach critical organs. The design incorporates key physical parameters and operates through particle reception and release mechanisms. In the reception process, described as ligand-receptor binding reactions, modeled as a continuous-time Markov process (CTMP), PaCoT uses metallothionein proteins as receptors and heavy metals (e.g., Zn, Pb, Cd) as ligands. We assume that the toxicity condition (toxic (bit-1), non-toxic (bit-0)) is encoded into the concentration of heavy metal molecules. Thus, we consider that heavy metal concentration within the MC channel (e.g., human circulatory system) employs binary concentration shift keying (binary CSK). The concentration ratio of specific heavy metals is estimated to infer toxicity, i.e., a high ratio indicates toxicity and a low ratio suggests non-toxicity. Toxicity detection is achieved by monitoring the receptor bound duration in the presence of interferers and various types of heavy metals. After detecting and collecting toxic heavy metals, PaCoT securely retains them in a liquid medium (e.g., water) until release, employing two mechanisms: (1) a single-disc viscous micropump to regulate flow rate, and (2) Brownian motion to facilitate diffusion. PaCoT's performance is evaluated through MATLAB simulations, focusing on bit error probability (BEP) of the toxicity detection method, release time of molecules from PaCoT and energy consumption.
2501.06099
Explaining Deep Learning-based Anomaly Detection in Energy Consumption Data by Focusing on Contextually Relevant Data
cs.LG cs.AI
Detecting anomalies in energy consumption data is crucial for identifying energy waste, equipment malfunction, and overall, for ensuring efficient energy management. Machine learning, and specifically deep learning approaches, have been greatly successful in anomaly detection; however, they are black-box approaches that do not provide transparency or explanations. SHAP and its variants have been proposed to explain these models, but they suffer from high computational complexity (SHAP) or instability and inconsistency (e.g., Kernel SHAP). To address these challenges, this paper proposes an explainability approach for anomalies in energy consumption data that focuses on context-relevant information. The proposed approach leverages existing explainability techniques, focusing on SHAP variants, together with global feature importance and weighted cosine similarity to select background dataset based on the context of each anomaly point. By focusing on the context and most relevant features, this approach mitigates the instability of explainability algorithms. Experimental results across 10 different machine learning models, five datasets, and five XAI techniques, demonstrate that our method reduces the variability of explanations providing consistent explanations. Statistical analyses confirm the robustness of our approach, showing an average reduction in variability of approximately 38% across multiple datasets.
2501.06100
Practical Quantum Circuit Implementation for Simulating Coupled Classical Oscillators
quant-ph cs.CE
Simulating large-scale coupled-oscillator systems presents substantial computational challenges for classical algorithms, particularly when pursuing first-principles analyses in the thermodynamic limit. Motivated by the quantum algorithm framework proposed by Babbush et al., we present and implement a detailed quantum circuit construction for simulating one-dimensional spring-mass systems. Our approach incorporates key quantum subroutines, including block encoding, quantum singular value transformation (QSVT), and amplitude amplification, to realize the unitary time-evolution operator associated with simulating classical oscillators dynamics. In the uniform spring-mass setting, our circuit construction requires a gate complexity of $\mathcal{O}\bigl(\log_2^2 N\,\log_2(1/\varepsilon)\bigr)$, where $N$ is the number of oscillators and $\varepsilon$ is the target accuracy of the approximation. For more general, heterogeneous spring-mass systems, the total gate complexity is $\mathcal{O}\bigl(N\log_2 N\,\log_2(1/\varepsilon)\bigr)$. Both settings require $\mathcal{O}(\log_2 N)$ qubits. Numerical simulations agree with classical solvers across all tested configurations, indicating that this circuit-based Hamiltonian simulation approach can substantially reduce computational costs and potentially enable larger-scale many-body studies on future quantum hardware.
2501.06101
From Conversation to Automation: Leveraging LLMs for Problem-Solving Therapy Analysis
cs.CL
Problem-solving therapy (PST) is a structured psychological approach that helps individuals manage stress and resolve personal issues by guiding them through problem identification, solution brainstorming, decision-making, and outcome evaluation. As mental health care increasingly adopts technologies like chatbots and large language models (LLMs), it is important to thoroughly understand how each session of PST is conducted before attempting to automate it. We developed a comprehensive framework for PST annotation using established PST Core Strategies and a set of novel Facilitative Strategies to analyze a corpus of real-world therapy transcripts to determine which strategies are most prevalent. Using various LLMs and transformer-based models, we found that GPT-4o outperformed all models, achieving the highest accuracy (0.76) in identifying all strategies. To gain deeper insights, we examined how strategies are applied by analyzing Therapeutic Dynamics (autonomy, self-disclosure, and metaphor), and linguistic patterns within our labeled data. Our research highlights LLMs' potential to automate therapy dialogue analysis, offering a scalable tool for mental health interventions. Our framework enhances PST by improving accessibility, effectiveness, and personalized support for therapists.
2501.06103
Finite-Horizon Single-Pull Restless Bandits: An Efficient Index Policy For Scarce Resource Allocation
cs.MA cs.LG
Restless multi-armed bandits (RMABs) have been highly successful in optimizing sequential resource allocation across many domains. However, in many practical settings with highly scarce resources, where each agent can only receive at most one resource, such as healthcare intervention programs, the standard RMAB framework falls short. To tackle such scenarios, we introduce Finite-Horizon Single-Pull RMABs (SPRMABs), a novel variant in which each arm can only be pulled once. This single-pull constraint introduces additional complexity, rendering many existing RMAB solutions suboptimal or ineffective. %To address this, we propose using dummy states to duplicate the system, ensuring that once an arm is activated, it transitions exclusively within the dummy states. To address this shortcoming, we propose using \textit{dummy states} that expand the system and enforce the one-pull constraint. We then design a lightweight index policy for this expanded system. For the first time, we demonstrate that our index policy achieves a sub-linearly decaying average optimality gap of $\tilde{\mathcal{O}}\left(\frac{1}{\rho^{1/2}}\right)$ for a finite number of arms, where $\rho$ is the scaling factor for each arm cluster. Extensive simulations validate the proposed method, showing robust performance across various domains compared to existing benchmarks.
2501.06104
Weather-Driven Priority Charging for Battery Storage Systems in Hybrid Renewable Energy Grids
eess.SY cs.SY
The integration of renewable energy into the power grid is often hindered by its fragmented infrastructure, leading to inefficient utilization due to the variability of energy production and its reliance on weather conditions. Battery storage systems, while essential for stabilizing energy supply, face challenges like sub-optimal energy distribution, accelerating battery degradation, and reducing operational efficiency. This paper presents a novel solution to these challenges by developing a large-scale, interconnected renewable energy network that optimizes energy storage and distribution. The proposed system includes strategically placed battery storage facilities that stabilize energy production by compensating for fluctuations in renewable output. A priority charging algorithm, informed by real-time weather forecasting and load monitoring, ensures that the most suitable battery systems are charged under varying conditions. Within each storage facility, a secondary priority charging algorithm minimizes battery degradation by ranking batteries based on critical parameters such as state of health (SoH) and state of charge (SoC) and deciding which to charge. This comprehensive approach enhances the efficiency and longevity of battery storage systems, offering a more reliable and resilient renewable energy infrastructure.
2501.06108
Inferring High-Order Couplings with Neural Networks
cond-mat.dis-nn cond-mat.stat-mech cs.LG
Maximum entropy methods, based on the inverse Ising/Potts problem from statistical mechanics, are essential for modeling interactions between pairs of variables in data-driven problems across disciplines such as bioinformatics, ecology, and neuroscience. Despite their considerable success, these methods typically fail to capture higher-order interactions that are often essential for understanding complex systems. Conversely, modern machine learning methods capture these complex interactions, but the computational cost of interpretable frameworks makes them impractical for real-world applications. Restricted Boltzmann Machines (RBMs) provide a computationally efficient way to capture statistical correlations using hidden nodes in a bipartite neural network. In this study, we introduce a new method that maps RBMs to generalized Potts models, allowing for the extraction of interactions up to any specified order. This method utilizes large-$N$ approximations, enabled by the RBM's simple structure, to extract effective many-body couplings with minimal computational effort. Furthermore, we propose a robust framework for extracting higher-order interactions in more complex probabilistic models and a simple gauge-fixing method within the effective many-body Potts model. Our validation on synthetic datasets confirms the method's ability to recover two- and three-body interactions accurately. When applied to protein sequence data, the framework competently reconstructs protein contact maps and provides performance comparable to the best inverse Potts models. These findings confirm that RBMs are an effective and streamlined tool for exploring higher-order interactions within complex systems.
2501.06112
Optimizing Experiments for Accurate Battery Circuit Parameters Estimation: Reduction and Adjustment of Frequency Set Used in Electrochemical Impedance Spectroscopy
eess.SY cs.SY
In this paper, we study a suitable experimental design of electrochemical impedance spectroscopy (EIS) to reduce the number of frequency points while not significantly affecting the uncertainties of the estimated cell's equivalent circuit model (ECM) parameters. It is based on an E-optimal experimental design that aims to maximize the information about the ECM parameters collected by EIS measurements and, at the same time, minimize the overall uncertainty. In a numerical experiment, we first analyze to which extent reducing the number of measurement points at low frequencies affects the uncertainty of the estimated parameters. Secondly, we show that applying the frequency adjustments can lead to the same or even improved global uncertainty of ECM parameter estimates as with a higher number of measurements. This is numerically verified through a case study using the ECM parameters of a commercial battery cell.
2501.06113
Vehicle-in-Virtual-Environment (VVE) Based Autonomous Driving Function Development and Evaluation Methodology for Vulnerable Road User Safety
cs.RO cs.SY eess.SY
Traditional methods for developing and evaluating autonomous driving functions, such as model-in-the-loop (MIL) and hardware-in-the-loop (HIL) simulations, heavily depend on the accuracy of simulated vehicle models and human factors, especially for vulnerable road user safety systems. Continuation of development during public road deployment forces other road users including vulnerable ones to involuntarily participate in the development process, leading to safety risks, inefficiencies, and a decline in public trust. To address these deficiencies, the Vehicle-in-Virtual-Environment (VVE) method was proposed as a safer, more efficient, and cost-effective solution for developing and testing connected and autonomous driving technologies by operating the real vehicle and multiple other actors like vulnerable road users in different test areas while being immersed within the same highly realistic virtual environment. This VVE approach synchronizes real-world vehicle and vulnerable road user motion within the same virtual scenario, enabling the safe and realistic testing of various traffic situations in a safe and repeatable manner. In this paper, we propose a new testing pipeline that sequentially integrates MIL, HIL, and VVE methods to comprehensively develop and evaluate autonomous driving functions. The effectiveness of this testing pipeline will be demonstrated using an autonomous driving path-tracking algorithm with local deep reinforcement learning modification for vulnerable road user collision avoidance.
2501.06115
Development of an Advisory System for Parking of a Car and Trailer
cs.RO cs.SY eess.SY
Trailer parking is a challenging task due to the unstable nature of the vehicle-trailer system in reverse motion and the unintuitive steering actions required at the vehicle to accomplish the parking maneuver. This paper presents a strategy to tackle this kind of maneuver with an advisory graphic aid to help the human driver with the task of manually backing up the vehicle-trailer system. A kinematic vehicle-trailer model is derived to describe the low-speed motion of the vehicle-trailer system, and its inverse kinematics is established by generating an equivalent virtual trailer axle steering command. The advisory system graphics is generated based on the inverse kinematics and displays the expected trailer orientation given the current vehicle steer angle and configuration (hitch angle). Simulation study and animation are set up to test the efficacy of the approach, where the user can select both vehicle speed and vehicle steering angle freely, which allows the user to stop the vehicle-trailer system and experiment with different steering inputs to see their effect on the predicted trailer motion before proceeding with the best one according to the advisory graphics, hence creating a series of piecewise continuous control actions similar to how manual trailer reverse parking is usually carried out. The advisory graphics proves to provide the driver with an intuitive understanding of the trailer motion at any given configuration (hitch angle).
2501.06117
Fleurs-SLU: A Massively Multilingual Benchmark for Spoken Language Understanding
cs.CL cs.AI
Spoken language understanding (SLU) is indispensable for half of all living languages that lack a formal writing system, since these languages cannot pair automatic speech recognition (ASR) with language models to benefit from language technology. Even if low-resource languages possess a writing system, ASR for these languages remains unreliable due to limited bimodal speech and text training data. Better SLU can strengthen the robustness of massively multilingual ASR by levering language semantics to disambiguate utterances via context or exploiting semantic similarities across languages. However, the evaluation of multilingual SLU remains limited to shallow tasks such as intent classification or language identification. To address this, we present Fleurs-SLU, a multilingual SLU benchmark that encompasses (i) 692 hours of speech for topical utterance classification in 102 languages and (ii) multiple-choice question answering through listening comprehension spanning 944 hours of speech across 92 languages. We extensively evaluate both end-to-end speech classification models and cascaded systems that combine speech-to-text transcription with subsequent classification by large language models on Fleurs-SLU. Our results show that cascaded systems exhibit greater robustness in multilingual SLU tasks, though speech encoders can achieve competitive performance in topical speech classification when appropriately pre-trained. We further find a strong correlation between robust multilingual ASR, effective speech-to-text translation, and strong multilingual SLU, highlighting the mutual benefits between acoustic and semantic speech representations.
2501.06118
Nonlinear port-Hamiltonian system identification from input-state-output data
eess.SY cs.SY math.DS math.OC nlin.CD
A framework for identifying nonlinear port-Hamiltonian systems using input-state-output data is introduced. The framework utilizes neural networks' universal approximation capacity to effectively represent complex dynamics in a structured way. We show that using the structure helps to make long-term predictions compared to baselines that do not incorporate physics. We also explore different architectures based on MLPs, KANs, and using prior information. The technique is validated through examples featuring nonlinearities in either the skew-symmetric terms, the dissipative terms, or the Hamiltonian.
2501.06121
kANNolo: Sweet and Smooth Approximate k-Nearest Neighbors Search
cs.IR
Approximate Nearest Neighbors (ANN) search is a crucial task in several applications like recommender systems and information retrieval. Current state-of-the-art ANN libraries, although being performance-oriented, often lack modularity and ease of use. This translates into them not being fully suitable for easy prototyping and testing of research ideas, an important feature to enable. We address these limitations by introducing kANNolo, a novel research-oriented ANN library written in Rust and explicitly designed to combine usability with performance effectively. kANNolo is the first ANN library that supports dense and sparse vector representations made available on top of different similarity measures, e.g., euclidean distance and inner product. Moreover, it also supports vector quantization techniques, e.g., Product Quantization, on top of the indexing strategies implemented. These functionalities are managed through Rust traits, allowing shared behaviors to be handled abstractly. This abstraction ensures flexibility and facilitates an easy integration of new components. In this work, we detail the architecture of kANNolo and demonstrate that its flexibility does not compromise performance. The experimental analysis shows that kANNolo achieves state-of-the-art performance in terms of speed-accuracy trade-off while allowing fast and easy prototyping, thus making kANNolo a valuable tool for advancing ANN research. Source code available on GitHub: https://github.com/TusKANNy/kannolo.
2501.06122
NDOB-Based Control of a UAV with Delta-Arm Considering Manipulator Dynamics
cs.RO
Aerial Manipulators (AMs) provide a versatile platform for various applications, including 3D printing, architecture, and aerial grasping missions. However, their operational speed is often sacrificed to uphold precision. Existing control strategies for AMs often regard the manipulator as a disturbance and employ robust control methods to mitigate its influence. This research focuses on elevating the precision of the end-effector and enhancing the agility of aerial manipulator movements. We present a composite control scheme to address these challenges. Initially, a Nonlinear Disturbance Observer (NDOB) is utilized to compensate for internal coupling effects and external disturbances. Subsequently, manipulator dynamics are processed through a high pass filter to facilitate agile movements. By integrating the proposed control method into a fully autonomous delta-arm-based AM system, we substantiate the controller's efficacy through extensive real-world experiments. The outcomes illustrate that the end-effector can achieve accuracy at the millimeter level.
2501.06126
Merging Feed-Forward Sublayers for Compressed Transformers
cs.CL cs.LG
With the rise and ubiquity of larger deep learning models, the need for high-quality compression techniques is growing in order to deploy these models widely. The sheer parameter count of these models makes it difficult to fit them into the memory constraints of different hardware. In this work, we present a novel approach to model compression by merging similar parameter groups within a model, rather than pruning away less important parameters. Specifically, we select, align, and merge separate feed-forward sublayers in Transformer models, and test our method on language modeling, image classification, and machine translation. With our method, we demonstrate performance comparable to the original models while combining more than a third of model feed-forward sublayers, and demonstrate improved performance over a strong layer-pruning baseline. For instance, we can remove over 21% of total parameters from a Vision Transformer, while maintaining 99% of its original performance. Additionally, we observe that some groups of feed-forward sublayers exhibit high activation similarity, which may help explain their surprising mergeability.
2501.06129
Contextual ASR Error Handling with LLMs Augmentation for Goal-Oriented Conversational AI
cs.CL cs.AI
General-purpose automatic speech recognition (ASR) systems do not always perform well in goal-oriented dialogue. Existing ASR correction methods rely on prior user data or named entities. We extend correction to tasks that have no prior user data and exhibit linguistic flexibility such as lexical and syntactic variations. We propose a novel context augmentation with a large language model and a ranking strategy that incorporates contextual information from the dialogue states of a goal-oriented conversational AI and its tasks. Our method ranks (1) n-best ASR hypotheses by their lexical and semantic similarity with context and (2) context by phonetic correspondence with ASR hypotheses. Evaluated in home improvement and cooking domains with real-world users, our method improves recall and F1 of correction by 34% and 16%, respectively, while maintaining precision and false positive rate. Users rated .8-1 point (out of 5) higher when our correction method worked properly, with no decrease due to false positives.
2501.06130
A Mixed-Integer Conic Program for the Multi-Agent Moving-Target Traveling Salesman Problem
cs.RO
The Moving-Target Traveling Salesman Problem (MT-TSP) aims to find a shortest path for an agent that starts at a stationary depot, visits a set of moving targets exactly once, each within one of their respective time windows, and then returns to the depot. In this paper, we introduce a new Mixed-Integer Conic Program (MICP) formulation that finds the optimum for the Multi-Agent Moving-Target Traveling Salesman Problem (MA-MT-TSP), a generalization of the MT-TSP involving multiple agents. We obtain our formulation by first restating the current state-of-the-art MICP formulation for MA-MT-TSP as a Mixed-Integer Nonlinear Nonconvex Program, and then reformulating it as a new MICP. We present computational results to demonstrate the performance of our approach. The results show that our formulation significantly outperforms the state-of-the-art, with up to a two-order-of-magnitude reduction in runtime, and up to over 90% tighter optimality gap.
2501.06132
CoDriveVLM: VLM-Enhanced Urban Cooperative Dispatching and Motion Planning for Future Autonomous Mobility on Demand Systems
cs.RO cs.AI cs.MA
The increasing demand for flexible and efficient urban transportation solutions has spotlighted the limitations of traditional Demand Responsive Transport (DRT) systems, particularly in accommodating diverse passenger needs and dynamic urban environments. Autonomous Mobility-on-Demand (AMoD) systems have emerged as a promising alternative, leveraging connected and autonomous vehicles (CAVs) to provide responsive and adaptable services. However, existing methods primarily focus on either vehicle scheduling or path planning, which often simplify complex urban layouts and neglect the necessity for simultaneous coordination and mutual avoidance among CAVs. This oversimplification poses significant challenges to the deployment of AMoD systems in real-world scenarios. To address these gaps, we propose CoDriveVLM, a novel framework that integrates high-fidelity simultaneous dispatching and cooperative motion planning for future AMoD systems. Our method harnesses Vision-Language Models (VLMs) to enhance multi-modality information processing, and this enables comprehensive dispatching and collision risk evaluation. The VLM-enhanced CAV dispatching coordinator is introduced to effectively manage complex and unforeseen AMoD conditions, thus supporting efficient scheduling decision-making. Furthermore, we propose a scalable decentralized cooperative motion planning method via consensus alternating direction method of multipliers (ADMM) focusing on collision risk evaluation and decentralized trajectory optimization. Simulation results demonstrate the feasibility and robustness of CoDriveVLM in various traffic conditions, showcasing its potential to significantly improve the fidelity and effectiveness of AMoD systems in future urban transportation networks. The code is available at https://github.com/henryhcliu/CoDriveVLM.git.
2501.06137
Supervision policies can shape long-term risk management in general-purpose AI models
cs.AI cs.CY cs.SI
The rapid proliferation and deployment of General-Purpose AI (GPAI) models, including large language models (LLMs), present unprecedented challenges for AI supervisory entities. We hypothesize that these entities will need to navigate an emergent ecosystem of risk and incident reporting, likely to exceed their supervision capacity. To investigate this, we develop a simulation framework parameterized by features extracted from the diverse landscape of risk, incident, or hazard reporting ecosystems, including community-driven platforms, crowdsourcing initiatives, and expert assessments. We evaluate four supervision policies: non-prioritized (first-come, first-served), random selection, priority-based (addressing the highest-priority risks first), and diversity-prioritized (balancing high-priority risks with comprehensive coverage across risk types). Our results indicate that while priority-based and diversity-prioritized policies are more effective at mitigating high-impact risks, particularly those identified by experts, they may inadvertently neglect systemic issues reported by the broader community. This oversight can create feedback loops that amplify certain types of reporting while discouraging others, leading to a skewed perception of the overall risk landscape. We validate our simulation results with several real-world datasets, including one with over a million ChatGPT interactions, of which more than 150,000 conversations were identified as risky. This validation underscores the complex trade-offs inherent in AI risk supervision and highlights how the choice of risk management policies can shape the future landscape of AI risks across diverse GPAI models used in society.
2501.06138
MS-Temba : Multi-Scale Temporal Mamba for Efficient Temporal Action Detection
cs.CV
Action detection in real-world scenarios is particularly challenging due to densely distributed actions in hour-long untrimmed videos. It requires modeling both short- and long-term temporal relationships while handling significant intra-class temporal variations. Previous state-of-the-art (SOTA) Transformer-based architectures, though effective, are impractical for real-world deployment due to their high parameter count, GPU memory usage, and limited throughput, making them unsuitable for very long videos. In this work, we innovatively adapt the Mamba architecture for action detection and propose Multi-scale Temporal Mamba (MS-Temba), comprising two key components: Temporal Mamba (Temba) Blocks and the Temporal Mamba Fuser. Temba Blocks include the Temporal Local Module (TLM) for short-range temporal modeling and the Dilated Temporal SSM (DTS) for long-range dependencies. By introducing dilations, a novel concept for Mamba, TLM and DTS capture local and global features at multiple scales. The Temba Fuser aggregates these scale-specific features using Mamba to learn comprehensive multi-scale representations of untrimmed videos. MS-Temba is validated on three public datasets, outperforming SOTA methods on long videos and matching prior methods on short videos while using only one-eighth of the parameters.
2501.06141
Emergent Symbol-like Number Variables in Artificial Neural Networks
cs.LG cs.AI cs.SC
What types of numeric representations emerge in Neural Networks (NNs)? To what degree do NNs induce abstract, mutable, slot-like numeric variables, and in what situations do these representations emerge? How do these representations change over learning, and how can we understand the neural implementations in ways that are unified across different NNs? In this work, we approach these questions by first training sequence based neural systems using Next Token Prediction (NTP) objectives on numeric tasks. We then seek to understand the neural solutions through the lens of causal abstractions or symbolic algorithms. We use a combination of causal interventions and visualization methods to find that artificial neural models do indeed develop analogs of interchangeable, mutable, latent number variables purely from the NTP objective. We then ask how variations on the tasks and model architectures affect the models' learned solutions to find that these symbol-like numeric representations do not form for every variant of the task, and transformers solve the problem in a notably different way than their recurrent counterparts. We then show how the symbol-like variables change over the course of training to find a strong correlation between the models' task performance and the alignment of their symbol-like representations. Lastly, we show that in all cases, some degree of gradience exists in these neural symbols, highlighting the difficulty of finding simple, interpretable symbolic stories of how neural networks perform numeric tasks. Taken together, our results are consistent with the view that neural networks can approximate interpretable symbolic programs of number cognition, but the particular program they approximate and the extent to which they approximate it can vary widely, depending on the network architecture, training data, extent of training, and network size.
2501.06143
Multilingual Performance of a Multimodal Artificial Intelligence System on Multisubject Physics Concept Inventories
physics.ed-ph cs.AI
We investigate the multilingual and multimodal performance of a large language model-based artificial intelligence (AI) system, GPT-4o, on a diverse set of physics concept inventories spanning multiple languages and subject areas. The inventories taken from the PhysPort website cover the classical physics topics of mechanics, electromagnetism, optics, and thermodynamics as well as relativity, quantum mechanics, astronomy, mathematics, and laboratory skills. Unlike previous text-only studies, we uploaded the inventories as images mirroring what a student would see on paper, assessing the system's multimodal functionality. The AI is prompted in English and autonomously chooses the language of its response - either remaining in the nominal language of the test, switching entirely to English, or mixing languages - revealing adaptive behavior dependent on linguistic complexity and data availability. Our results indicate some variation in performance across subject areas, with laboratory skills standing out as the area of poorest performance. Furthermore, the AI's performance on questions that require visual interpretation of images is worse than on purely text-based questions. Questions that are difficult for the AI tend to be that way invariably of the inventory language. We also find large variations in performance across languages, with some appearing to benefit substantially from language switching, a phenomenon similar to code-switching ofhuman speakers. Overall, comparing the obtained AI results to the existing literature, we find that the AI system outperforms average undergraduate students post-instruction in all subject areas but laboratory skills.
2501.06146
xLSTM-SENet: xLSTM for Single-Channel Speech Enhancement
cs.SD cs.AI eess.AS
While attention-based architectures, such as Conformers, excel in speech enhancement, they face challenges such as scalability with respect to input sequence length. In contrast, the recently proposed Extended Long Short-Term Memory (xLSTM) architecture offers linear scalability. However, xLSTM-based models remain unexplored for speech enhancement. This paper introduces xLSTM-SENet, the first xLSTM-based single-channel speech enhancement system. A comparative analysis reveals that xLSTM-and notably, even LSTM-can match or outperform state-of-the-art Mamba- and Conformer-based systems across various model sizes in speech enhancement on the VoiceBank+Demand dataset. Through ablation studies, we identify key architectural design choices such as exponential gating and bidirectionality contributing to its effectiveness. Our best xLSTM-based model, xLSTM-SENet2, outperforms state-of-the-art Mamba- and Conformer-based systems on the Voicebank+DEMAND dataset.
2501.06148
From discrete-time policies to continuous-time diffusion samplers: Asymptotic equivalences and faster training
cs.LG stat.ML
We study the problem of training neural stochastic differential equations, or diffusion models, to sample from a Boltzmann distribution without access to target samples. Existing methods for training such models enforce time-reversal of the generative and noising processes, using either differentiable simulation or off-policy reinforcement learning (RL). We prove equivalences between families of objectives in the limit of infinitesimal discretization steps, linking entropic RL methods (GFlowNets) with continuous-time objects (partial differential equations and path space measures). We further show that an appropriate choice of coarse time discretization during training allows greatly improved sample efficiency and the use of time-local objectives, achieving competitive performance on standard sampling benchmarks with reduced computational cost.
2501.06151
PySpatial: A High-Speed Whole Slide Image Pathomics Toolkit
eess.IV cs.CV
Whole Slide Image (WSI) analysis plays a crucial role in modern digital pathology, enabling large-scale feature extraction from tissue samples. However, traditional feature extraction pipelines based on tools like CellProfiler often involve lengthy workflows, requiring WSI segmentation into patches, feature extraction at the patch level, and subsequent mapping back to the original WSI. To address these challenges, we present PySpatial, a high-speed pathomics toolkit specifically designed for WSI-level analysis. PySpatial streamlines the conventional pipeline by directly operating on computational regions of interest, reducing redundant processing steps. Utilizing rtree-based spatial indexing and matrix-based computation, PySpatial efficiently maps and processes computational regions, significantly accelerating feature extraction while maintaining high accuracy. Our experiments on two datasets-Perivascular Epithelioid Cell (PEC) and data from the Kidney Precision Medicine Project (KPMP)-demonstrate substantial performance improvements. For smaller and sparse objects in PEC datasets, PySpatial achieves nearly a 10-fold speedup compared to standard CellProfiler pipelines. For larger objects, such as glomeruli and arteries in KPMP datasets, PySpatial achieves a 2-fold speedup. These results highlight PySpatial's potential to handle large-scale WSI analysis with enhanced efficiency and accuracy, paving the way for broader applications in digital pathology.
2501.06158
GenMol: A Drug Discovery Generalist with Discrete Diffusion
cs.LG
Drug discovery is a complex process that involves multiple scenarios and stages, such as fragment-constrained molecule generation, hit generation and lead optimization. However, existing molecular generative models can only tackle one or two of these scenarios and lack the flexibility to address various aspects of the drug discovery pipeline. In this paper, we present Generalist Molecular generative model (GenMol), a versatile framework that addresses these limitations by applying discrete diffusion to the Sequential Attachment-based Fragment Embedding (SAFE) molecular representation. GenMol generates SAFE sequences through non-autoregressive bidirectional parallel decoding, thereby allowing utilization of a molecular context that does not rely on the specific token ordering and enhanced computational efficiency. Moreover, under the discrete diffusion framework, we introduce fragment remasking, a strategy that optimizes molecules by replacing fragments with masked tokens and regenerating them, enabling effective exploration of chemical space. GenMol significantly outperforms the previous GPT-based model trained on SAFE representations in de novo generation and fragment-constrained generation, and achieves state-of-the-art performance in goal-directed hit generation and lead optimization. These experimental results demonstrate that GenMol can tackle a wide range of drug discovery tasks, providing a unified and versatile approach for molecular design.
2501.06159
Efficient Transition State Searches by Freezing String Method with Graph Neural Network Potentials
physics.chem-ph cs.LG
Transition states are a critical bottleneck in chemical transformations. Significant efforts have been made to develop algorithms that efficiently locate transition states on potential energy surfaces. However, the computational cost of ab-initio potential energy surface evaluation limits the size of chemical systems that can routinely studied. In this work, we develop and fine-tune a graph neural network potential energy function suitable for describing organic chemical reactions and use it to rapidly identify transition state guess structures. We successfully refine guess structures and locate a transition state in each test system considered and reduce the average number of ab-initio calculations by 47% though use of the graph neural network potential energy function. Our results show that modern machine learning models have reached levels of reliability whereby they can be used to accelerate routine computational chemistry tasks.
2501.06164
Model Alignment Search
cs.LG cs.AI
When can we say that two neural systems are the same? The answer to this question is goal-dependent, and it is often addressed through correlative methods such as Representational Similarity Analysis (RSA) and Centered Kernel Alignment (CKA). How do we target functionally relevant similarity, and how do we isolate specific causal aspects of the representations? In this work, we introduce Model Alignment Search (MAS), a method for causally exploring distributed representational similarity. The method learns invertible linear transformations that align a subspace between two distributed networks' representations where causal information can be freely interchanged. We first show that the method can be used to transfer values of specific causal variables -- such as the number of items in a counting task -- between networks with different training seeds. We then explore open questions in number cognition by comparing different types of numeric representations in models trained on structurally different tasks. We then explore differences between MAS vs preexisting causal similarity methods, and lastly, we introduce a counterfactual latent auxiliary loss function that helps shape causally relevant alignments even in cases where we do not have causal access to one of the two models for training.
2501.06167
Meta-Learning for Physically-Constrained Neural System Identification
cs.LG cs.SY eess.SY math.OC
We present a gradient-based meta-learning framework for rapid adaptation of neural state-space models (NSSMs) for black-box system identification. When applicable, we also incorporate domain-specific physical constraints to improve the accuracy of the NSSM. The major benefit of our approach is that instead of relying solely on data from a single target system, our framework utilizes data from a diverse set of source systems, enabling learning from limited target data, as well as with few online training iterations. Through benchmark examples, we demonstrate the potential of our approach, study the effect of fine-tuning subnetworks rather than full fine-tuning, and report real-world case studies to illustrate the practical application and generalizability of the approach to practical problems with physical-constraints. Specifically, we show that the meta-learned models result in improved downstream performance in model-based state estimation in indoor localization and energy systems.
2501.06171
Machine Learning Force-Field Approach for Itinerant Electron Magnets
cond-mat.str-el cs.LG physics.comp-ph
We review the recent development of machine-learning (ML) force-field frameworks for Landau-Lifshitz-Gilbert (LLG) dynamics simulations of itinerant electron magnets, focusing on the general theory and implementations of symmetry-invariant representations of spin configurations. The crucial properties that such magnetic descriptors must satisfy are differentiability with respect to spin rotations and invariance to both lattice point-group symmetry and internal spin rotation symmetry. We propose an efficient implementation based on the concept of reference irreducible representations, modified from the group-theoretical power-spectrum and bispectrum methods. The ML framework is demonstrated using the s-d models, which are widely applied in spintronics research. We show that LLG simulations based on local fields predicted by the trained ML models successfully reproduce representative non-collinear spin structures, including 120$^\circ$, tetrahedral, and skyrmion crystal orders of the triangular-lattice s-d models. Large-scale thermal quench simulations enabled by ML models further reveal intriguing freezing dynamics and glassy stripe states consisting of skyrmions and bi-merons. Our work highlights the utility of ML force-field approach to dynamical modeling of complex spin orders in itinerant electron magnets.
2501.06173
VideoAuteur: Towards Long Narrative Video Generation
cs.CV
Recent video generation models have shown promising results in producing high-quality video clips lasting several seconds. However, these models face challenges in generating long sequences that convey clear and informative events, limiting their ability to support coherent narrations. In this paper, we present a large-scale cooking video dataset designed to advance long-form narrative generation in the cooking domain. We validate the quality of our proposed dataset in terms of visual fidelity and textual caption accuracy using state-of-the-art Vision-Language Models (VLMs) and video generation models, respectively. We further introduce a Long Narrative Video Director to enhance both visual and semantic coherence in generated videos and emphasize the role of aligning visual embeddings to achieve improved overall video quality. Our method demonstrates substantial improvements in generating visually detailed and semantically aligned keyframes, supported by finetuning techniques that integrate text and image embeddings within the video generation process. Project page: https://videoauteur.github.io/
2501.06181
Best Response Convergence for Zero-sum Stochastic Dynamic Games with Partial and Asymmetric Information
eess.SY cs.SY math.OC
We analyze best response dynamics for finding a Nash equilibrium of an infinite horizon zero-sum stochastic linear quadratic dynamic game (LQDG) with partial and asymmetric information. We derive explicit expressions for each player's best response within the class of pure linear dynamic output feedback control strategies where the internal state dimension of each control strategy is an integer multiple of the system state dimension. With each best response, the players form increasingly higher-order belief states, leading to infinite-dimensional internal states. However, we observe in extensive numerical experiments that the game's value converges after just a few iterations, suggesting that strategies associated with increasingly higher-order belief states eventually provide no benefit. To help explain this convergence, our numerical analysis reveals rapid decay of the controllability and observability Gramian eigenvalues and Hankel singular values in higher-order belief dynamics, indicating that the higher-order belief dynamics become increasingly difficult for both players to control and observe. Consequently, the higher-order belief dynamics can be closely approximated by low-order belief dynamics with bounded error, and thus feedback strategies with limited internal state dimension can closely approximate a Nash equilibrium.
2501.06184
PEACE: Empowering Geologic Map Holistic Understanding with MLLMs
cs.CV cs.MA
Geologic map, as a fundamental diagram in geology science, provides critical insights into the structure and composition of Earth's subsurface and surface. These maps are indispensable in various fields, including disaster detection, resource exploration, and civil engineering. Despite their significance, current Multimodal Large Language Models (MLLMs) often fall short in geologic map understanding. This gap is primarily due to the challenging nature of cartographic generalization, which involves handling high-resolution map, managing multiple associated components, and requiring domain-specific knowledge. To quantify this gap, we construct GeoMap-Bench, the first-ever benchmark for evaluating MLLMs in geologic map understanding, which assesses the full-scale abilities in extracting, referring, grounding, reasoning, and analyzing. To bridge this gap, we introduce GeoMap-Agent, the inaugural agent designed for geologic map understanding, which features three modules: Hierarchical Information Extraction (HIE), Domain Knowledge Injection (DKI), and Prompt-enhanced Question Answering (PEQA). Inspired by the interdisciplinary collaboration among human scientists, an AI expert group acts as consultants, utilizing a diverse tool pool to comprehensively analyze questions. Through comprehensive experiments, GeoMap-Agent achieves an overall score of 0.811 on GeoMap-Bench, significantly outperforming 0.369 of GPT-4o. Our work, emPowering gEologic mAp holistiC undErstanding (PEACE) with MLLMs, paves the way for advanced AI applications in geology, enhancing the efficiency and accuracy of geological investigations.
2501.06186
LlamaV-o1: Rethinking Step-by-step Visual Reasoning in LLMs
cs.CV
Reasoning is a fundamental capability for solving complex multi-step problems, particularly in visual contexts where sequential step-wise understanding is essential. Existing approaches lack a comprehensive framework for evaluating visual reasoning and do not emphasize step-wise problem-solving. To this end, we propose a comprehensive framework for advancing step-by-step visual reasoning in large language models (LMMs) through three key contributions. First, we introduce a visual reasoning benchmark specifically designed to evaluate multi-step reasoning tasks. The benchmark presents a diverse set of challenges with eight different categories ranging from complex visual perception to scientific reasoning with over 4k reasoning steps in total, enabling robust evaluation of LLMs' abilities to perform accurate and interpretable visual reasoning across multiple steps. Second, we propose a novel metric that assesses visual reasoning quality at the granularity of individual steps, emphasizing both correctness and logical coherence. The proposed metric offers deeper insights into reasoning performance compared to traditional end-task accuracy metrics. Third, we present a new multimodal visual reasoning model, named LlamaV-o1, trained using a multi-step curriculum learning approach, where tasks are progressively organized to facilitate incremental skill acquisition and problem-solving. The proposed LlamaV-o1 is designed for multi-step reasoning and learns step-by-step through a structured training paradigm. Extensive experiments show that our LlamaV-o1 outperforms existing open-source models and performs favorably against close-source proprietary models. Compared to the recent Llava-CoT, our LlamaV-o1 achieves an average score of 67.3 with an absolute gain of 3.8\% across six benchmarks while being 5 times faster during inference scaling. Our benchmark, model, and code are publicly available.
2501.06187
Multi-subject Open-set Personalization in Video Generation
cs.CV
Video personalization methods allow us to synthesize videos with specific concepts such as people, pets, and places. However, existing methods often focus on limited domains, require time-consuming optimization per subject, or support only a single subject. We present Video Alchemist $-$ a video model with built-in multi-subject, open-set personalization capabilities for both foreground objects and background, eliminating the need for time-consuming test-time optimization. Our model is built on a new Diffusion Transformer module that fuses each conditional reference image and its corresponding subject-level text prompt with cross-attention layers. Developing such a large model presents two main challenges: dataset and evaluation. First, as paired datasets of reference images and videos are extremely hard to collect, we sample selected video frames as reference images and synthesize a clip of the target video. However, while models can easily denoise training videos given reference frames, they fail to generalize to new contexts. To mitigate this issue, we design a new automatic data construction pipeline with extensive image augmentations. Second, evaluating open-set video personalization is a challenge in itself. To address this, we introduce a personalization benchmark that focuses on accurate subject fidelity and supports diverse personalization scenarios. Finally, our extensive experiments show that our method significantly outperforms existing personalization methods in both quantitative and qualitative evaluations.
2501.06189
A Multimodal Social Agent
cs.AI cs.CL
In recent years, large language models (LLMs) have demonstrated remarkable progress in common-sense reasoning tasks. This ability is fundamental to understanding social dynamics, interactions, and communication. However, the potential of integrating computers with these social capabilities is still relatively unexplored. However, the potential of integrating computers with these social capabilities is still relatively unexplored. This paper introduces MuSA, a multimodal LLM-based agent that analyzes text-rich social content tailored to address selected human-centric content analysis tasks, such as question answering, visual question answering, title generation, and categorization. It uses planning, reasoning, acting, optimizing, criticizing, and refining strategies to complete a task. Our approach demonstrates that MuSA can automate and improve social content analysis, helping decision-making processes across various applications. We have evaluated our agent's capabilities in question answering, title generation, and content categorization tasks. MuSA performs substantially better than our baselines.
2501.06192
A Computational Model of Learning and Memory Using Structurally Dynamic Cellular Automata
cs.AI cs.NE math.DS q-bio.NC
In the fields of computation and neuroscience, much is still unknown about the underlying computations that enable key cognitive functions including learning, memory, abstraction and behavior. This paper proposes a mathematical and computational model of learning and memory based on a small set of bio-plausible functions that include coincidence detection, signal modulation, and reward/penalty mechanisms. Our theoretical approach proposes that these basic functions are sufficient to establish and modulate an information space over which computation can be carried out, generating signal gradients usable for inference and behavior. The computational method used to test this is a structurally dynamic cellular automaton with continuous-valued cell states and a series of recursive steps propagating over an undirected graph with the memory function embedded entirely in the creation and modulation of graph edges. The experimental results show: that the toy model can make near-optimal choices to re-discover a reward state after a single training run; that it can avoid complex penalty configurations; that signal modulation and network plasticity can generate exploratory behaviors in sparse reward environments; that the model generates context-dependent memory representations; and that it exhibits high computational efficiency because of its minimal, single-pass training requirements combined with flexible and contextual memory representation.
2501.06193
A Novel Task-Driven Method with Evolvable Interactive Agents Using Event Trees for Enhanced Emergency Decision Support
cs.AI cs.CL
As climate change and other global challenges increase the likelihood of unforeseen emergencies, the limitations of human-driven strategies in critical situations become more pronounced. Inadequate pre-established emergency plans can lead operators to become overwhelmed during complex systems malfunctions. This study addresses the urgent need for agile decision-making in response to various unforeseen incidents through a novel approach, EvoTaskTree (a task-driven method with evolvable interactive agents using event trees for emergency decision support). This advanced approach integrates two types of agents powered by large language models (LLMs): task executors, responsible for executing critical procedures, and task validators, ensuring the efficacy of those actions. By leveraging insights from event tree analysis, our framework encompasses three crucial tasks: initiating event subevent analysis, event tree header event analysis, and decision recommendations. The agents learn from both successful and unsuccessful responses from these tasks. Finally, we use nuclear power plants as a demonstration of a safety-critical system. Our findings indicate that the designed agents are not only effective but also outperform existing approaches, achieving an impressive accuracy rate of up to 100 % in processing previously unencoun32 tered incident scenarios. This paper demonstrates that EvoTaskTree significantly enhances the rapid formulation of emergency decision-making.
2501.06196
How Do Artificial Intelligences Think? The Three Mathematico-Cognitive Factors of Categorical Segmentation Operated by Synthetic Neurons
q-bio.NC cs.AI cs.NE
How do the synthetic neurons in language models create "thought categories" to segment and analyze their informational environment? What are the cognitive characteristics, at the very level of formal neurons, of this artificial categorical thought? Based on the mathematical nature of algebraic operations inherent to neuronal aggregation functions, we attempt to identify mathematico-cognitive factors that genetically shape the categorical reconstruction of the informational world faced by artificial cognition. This study explores these concepts through the notions of priming, attention, and categorical phasing.
2501.06201
A Novel Method for Pignistic Information Fusion in the View of Z-number
cs.AI
How to properly fuse information from complex sources is still an open problem. Lots of methods have been put forward to provide a effective solution in fusing intricate information. Among them, Dempster-Shafer evidences theory (DSET) is one of the representatives, it is widely used to handle uncertain information. Based on DSET, a completely new method to fuse information from different sources based on pignistic transformation and Z-numbers is proposed in this paper which is able to handle separate situations of information and keeps high accuracy in producing rational and correct judgments on actual situations. Besides, in order to illustrate the superiority of the proposed method, some numerical examples and application are also provided to verify the validity and robustness of it.
2501.06205
Leveraging Edge Intelligence and LLMs to Advance 6G-Enabled Internet of Automated Defense Vehicles
cs.NI cs.AI cs.CL
The evolution of Artificial Intelligence (AI) and its subset Deep Learning (DL), has profoundly impacted numerous domains, including autonomous driving. The integration of autonomous driving in military settings reduces human casualties and enables precise and safe execution of missions in hazardous environments while allowing for reliable logistics support without the risks associated with fatigue-related errors. However, relying on autonomous driving solely requires an advanced decision-making model that is adaptable and optimum in any situation. Considering the presence of numerous interconnected autonomous vehicles in mission-critical scenarios, Ultra-Reliable Low Latency Communication (URLLC) is vital for ensuring seamless coordination, real-time data exchange, and instantaneous response to dynamic driving environments. The advent of 6G strengthens the Internet of Automated Defense Vehicles (IoADV) concept within the realm of Internet of Military Defense Things (IoMDT) by enabling robust connectivity, crucial for real-time data exchange, advanced navigation, and enhanced safety features through IoADV interactions. On the other hand, a critical advancement in this space is using pre-trained Generative Large Language Models (LLMs) for decision-making and communication optimization for autonomous driving. Hence, this work presents opportunities and challenges with a vision of realizing the full potential of these technologies in critical defense applications, especially through the advancement of IoADV and its role in enhancing autonomous military operations.
2501.06208
Enhancing AI Safety Through the Fusion of Low Rank Adapters
cs.CL
Instruction fine-tuning of large language models (LLMs) is a powerful method for improving task-specific performance, but it can inadvertently lead to a phenomenon where models generate harmful responses when faced with malicious prompts. In this paper, we explore Low-Rank Adapter Fusion (LoRA) as a means to mitigate these risks while preserving the model's ability to handle diverse instructions effectively. Through an extensive comparative analysis against established baselines using recognized benchmark datasets, we demonstrate a 42\% reduction in the harmfulness rate by leveraging LoRA fusion between a task adapter and a safety adapter, the latter of which is specifically trained on our safety dataset. However, we also observe exaggerated safety behaviour, where the model rejects safe prompts that closely resemble unsafe ones
2501.06210
Applications of natural language processing in aviation safety: A review and qualitative analysis
cs.CL cs.LG
This study explores using Natural Language Processing in aviation safety, focusing on machine learning algorithms to enhance safety measures. There are currently May 2024, 34 Scopus results from the keyword search natural language processing and aviation safety. Analyzing these studies allows us to uncover trends in the methodologies, findings and implications of NLP in aviation. Both qualitative and quantitative tools have been used to investigate the current state of literature on NLP for aviation safety. The qualitative analysis summarises the research motivations, objectives, and outcomes, showing how NLP can be utilized to help identify critical safety issues and improve aviation safety. This study also identifies research gaps and suggests areas for future exploration, providing practical recommendations for the aviation industry. We discuss challenges in implementing NLP in aviation safety, such as the need for large, annotated datasets, and the difficulty in interpreting complex models. We propose solutions like active learning for data annotation and explainable AI for model interpretation. Case studies demonstrate the successful application of NLP in improving aviation safety, highlighting its potential to make aviation safer and more efficient.
2501.06211
FLAME: Financial Large-Language Model Assessment and Metrics Evaluation
cs.CL cs.AI cs.CE
LLMs have revolutionized NLP and demonstrated potential across diverse domains. More and more financial LLMs have been introduced for finance-specific tasks, yet comprehensively assessing their value is still challenging. In this paper, we introduce FLAME, a comprehensive financial LLMs evaluation system in Chinese, which includes two core evaluation benchmarks: FLAME-Cer and FLAME-Sce. FLAME-Cer covers 14 types of authoritative financial certifications, including CPA, CFA, and FRM, with a total of approximately 16,000 carefully selected questions. All questions have been manually reviewed to ensure accuracy and representativeness. FLAME-Sce consists of 10 primary core financial business scenarios, 21 secondary financial business scenarios, and a comprehensive evaluation set of nearly 100 tertiary financial application tasks. We evaluate 6 representative LLMs, including GPT-4o, GLM-4, ERNIE-4.0, Qwen2.5, XuanYuan3, and the latest Baichuan4-Finance, revealing Baichuan4-Finance excels other LLMs in most tasks. By establishing a comprehensive and professional evaluation system, FLAME facilitates the advancement of financial LLMs in Chinese contexts. Instructions for participating in the evaluation are available on GitHub: https://github.com/FLAME-ruc/FLAME.
2501.06214
Path Space Partitioning and Guided Image Sampling for MCMC
cs.CV cs.GR
Rendering algorithms typically integrate light paths over path space. However, integrating over this one unified space is not necessarily the most efficient approach, and we show that partitioning path space and integrating each of these partitioned spaces with a separate estimator can have advantages. We propose an approach for partitioning path space based on analyzing paths from a standard Monte Carlo estimator and integrating these partitioned path spaces using a Markov Chain Monte Carlo (MCMC) estimator. This also means that integration happens within a sparser subset of path space, so we propose the use of guided proposal distributions in image space to improve efficiency. We show that our method improves image quality over other MCMC integration approaches at the same number of samples.
2501.06215
Fitting Different Interactive Information: Joint Classification of Emotion and Intention
cs.CV cs.CL cs.LG cs.MM eess.AS
This paper is the first-place solution for ICASSP MEIJU@2025 Track I, which focuses on low-resource multimodal emotion and intention recognition. How to effectively utilize a large amount of unlabeled data, while ensuring the mutual promotion of different difficulty levels tasks in the interaction stage, these two points become the key to the competition. In this paper, pseudo-label labeling is carried out on the model trained with labeled data, and samples with high confidence and their labels are selected to alleviate the problem of low resources. At the same time, the characteristic of easy represented ability of intention recognition found in the experiment is used to make mutually promote with emotion recognition under different attention heads, and higher performance of intention recognition is achieved through fusion. Finally, under the refined processing data, we achieve the score of 0.5532 in the Test set, and win the championship of the track.
2501.06216
Understanding colors of Dufaycolor: Can we recover them using historical colorimetric and spectral data?
cs.CV cs.GR
Dufaycolor, an additive color photography process produced from 1935 to the late 1950s, represents one of the most advanced iterations of this technique. This paper presents ongoing research and development of an open-source Color-Screen tool designed to reconstruct the original colors of additive color photographs. We discuss the incorporation of historical measurements of dyes used in the production of the color-screen filter (r\'eseau) to achieve accurate color recovery.
2501.06218
Dissecting Bit-Level Scaling Laws in Quantizing Vision Generative Models
cs.CV
Vision generative models have recently made significant advancements along two primary paradigms: diffusion-style and language-style, both of which have demonstrated excellent scaling laws. Quantization is crucial for efficiently deploying these models, as it reduces memory and computation costs. In this work, we systematically investigate the impact of quantization on these two paradigms. Surprisingly, despite achieving comparable performance in full precision, language-style models consistently outperform diffusion-style models across various quantization settings. This observation suggests that language-style models have superior bit-level scaling laws, offering a better tradeoff between model quality and total bits. To dissect this phenomenon, we conduct extensive experiments and find that the primary reason is the discrete representation space of language-style models, which is more tolerant of information loss during quantization. Furthermore, our analysis indicates that improving the bit-level scaling law of quantized vision generative models is challenging, with model distillation identified as a highly effective approach. Specifically, we propose TopKLD to optimize the transfer of distilled knowledge by balancing ``implicit knowledge'' and ``explicit knowledge'' during the distillation process. This approach elevates the bit-level scaling laws by one level across both integer and floating-point quantization settings.
2501.06219
WhACC: Whisker Automatic Contact Classifier with Expert Human-Level Performance
cs.CV cs.LG
The rodent vibrissal system is pivotal in advancing neuroscience research, particularly for studies of cortical plasticity, learning, decision-making, sensory encoding, and sensorimotor integration. Despite the advantages, curating touch events is labor intensive and often requires >3 hours per million video frames, even after leveraging automated tools like the Janelia Whisker Tracker. We address this limitation by introducing Whisker Automatic Contact Classifier (WhACC), a python package designed to identify touch periods from high-speed videos of head-fixed behaving rodents with human-level performance. WhACC leverages ResNet50V2 for feature extraction, combined with LightGBM for Classification. Performance is assessed against three expert human curators on over one million frames. Pairwise touch classification agreement on 99.5% of video frames, equal to between-human agreement. Finally, we offer a custom retraining interface to allow model customization on a small subset of data, which was validated on four million frames across 16 single-unit electrophysiology recordings. Including this retraining step, we reduce human hours required to curate a 100 million frame dataset from ~333 hours to ~6 hours.
2501.06220
Powerful Design of Small Vision Transformer on CIFAR10
cs.LG cs.CV
Vision Transformers (ViTs) have demonstrated remarkable success on large-scale datasets, but their performance on smaller datasets often falls short of convolutional neural networks (CNNs). This paper explores the design and optimization of Tiny ViTs for small datasets, using CIFAR-10 as a benchmark. We systematically evaluate the impact of data augmentation, patch token initialization, low-rank compression, and multi-class token strategies on model performance. Our experiments reveal that low-rank compression of queries in Multi-Head Latent Attention (MLA) incurs minimal performance loss, indicating redundancy in ViTs. Additionally, introducing multiple CLS tokens improves global representation capacity, boosting accuracy. These findings provide a comprehensive framework for optimizing Tiny ViTs, offering practical insights for efficient and effective designs. Code is available at https://github.com/erow/PoorViTs.
2501.06221
Optimizing Supply Chain Networks with the Power of Graph Neural Networks
cs.LG econ.GN q-fin.EC
Graph Neural Networks (GNNs) have emerged as transformative tools for modeling complex relational data, offering unprecedented capabilities in tasks like forecasting and optimization. This study investigates the application of GNNs to demand forecasting within supply chain networks using the SupplyGraph dataset, a benchmark for graph-based supply chain analysis. By leveraging advanced GNN methodologies, we enhance the accuracy of forecasting models, uncover latent dependencies, and address temporal complexities inherent in supply chain operations. Comparative analyses demonstrate that GNN-based models significantly outperform traditional approaches, including Multilayer Perceptrons (MLPs) and Graph Convolutional Networks (GCNs), particularly in single-node demand forecasting tasks. The integration of graph representation learning with temporal data highlights GNNs' potential to revolutionize predictive capabilities for inventory management, production scheduling, and logistics optimization. This work underscores the pivotal role of forecasting in supply chain management and provides a robust framework for advancing research and applications in this domain.
2501.06222
Can Explainable AI Assess Personalized Health Risks from Indoor Air Pollution?
cs.LG
Acknowledging the effects of outdoor air pollution, the literature inadequately addresses indoor air pollution's impacts. Despite daily health risks, existing research primarily focused on monitoring, lacking accuracy in pinpointing indoor pollution sources. In our research work, we thoroughly investigated the influence of indoor activities on pollution levels. A survey of 143 participants revealed limited awareness of indoor air pollution. Leveraging 65 days of diverse data encompassing activities like incense stick usage, indoor smoking, inadequately ventilated cooking, excessive AC usage, and accidental paper burning, we developed a comprehensive monitoring system. We identify pollutant sources and effects with high precision through clustering analysis and interpretability models (LIME and SHAP). Our method integrates Decision Trees, Random Forest, Naive Bayes, and SVM models, excelling at 99.8% accuracy with Decision Trees. Continuous 24-hour data allows personalized assessments for targeted pollution reduction strategies, achieving 91% accuracy in predicting activities and pollution exposure.
2501.06223
Interpretable Auto Window Setting for Deep-Learning-Based CT Analysis
eess.IV cs.CV cs.LG
Whether during the early days of popularization or in the present, the window setting in Computed Tomography (CT) has always been an indispensable part of the CT analysis process. Although research has investigated the capabilities of CT multi-window fusion in enhancing neural networks, there remains a paucity of domain-invariant, intuitively interpretable methodologies for Auto Window Setting. In this work, we propose an plug-and-play module originate from Tanh activation function, which is compatible with mainstream deep learning architectures. Starting from the physical principles of CT, we adhere to the principle of interpretability to ensure the module's reliability for medical implementations. The domain-invariant design facilitates observation of the preference decisions rendered by the adaptive mechanism from a clinically intuitive perspective. This enables the proposed method to be understood not only by experts in neural networks but also garners higher trust from clinicians. We confirm the effectiveness of the proposed method in multiple open-source datasets, yielding 10%~200% Dice improvements on hard segment targets.
2501.06224
Detection, Retrieval, and Explanation Unified: A Violence Detection System Based on Knowledge Graphs and GAT
cs.CV cs.AI
Recently, violence detection systems developed using unified multimodal models have achieved significant success and attracted widespread attention. However, most of these systems face two critical challenges: the lack of interpretability as black-box models and limited functionality, offering only classification or retrieval capabilities. To address these challenges, this paper proposes a novel interpretable violence detection system, termed the Three-in-One (TIO) System. The TIO system integrates knowledge graphs (KG) and graph attention networks (GAT) to provide three core functionalities: detection, retrieval, and explanation. Specifically, the system processes each video frame along with text descriptions generated by a large language model (LLM) for videos containing potential violent behavior. It employs ImageBind to generate high-dimensional embeddings for constructing a knowledge graph, uses GAT for reasoning, and applies lightweight time series modules to extract video embedding features. The final step connects a classifier and retriever for multi-functional outputs. The interpretability of KG enables the system to verify the reasoning process behind each output. Additionally, the paper introduces several lightweight methods to reduce the resource consumption of the TIO system and enhance its efficiency. Extensive experiments conducted on the XD-Violence and UCF-Crime datasets validate the effectiveness of the proposed system. A case study further reveals an intriguing phenomenon: as the number of bystanders increases, the occurrence of violent behavior tends to decrease.
2501.06225
A Distributed Hybrid Quantum Convolutional Neural Network for Medical Image Classification
cs.CV cs.LG
Medical images are characterized by intricate and complex features, requiring interpretation by physicians with medical knowledge and experience. Classical neural networks can reduce the workload of physicians, but can only handle these complex features to a limited extent. Theoretically, quantum computing can explore a broader parameter space with fewer parameters, but it is currently limited by the constraints of quantum hardware.Considering these factors, we propose a distributed hybrid quantum convolutional neural network based on quantum circuit splitting. This model leverages the advantages of quantum computing to effectively capture the complex features of medical images, enabling efficient classification even in resource-constrained environments. Our model employs a quantum convolutional neural network (QCNN) to extract high-dimensional features from medical images, thereby enhancing the model's expressive capability.By integrating distributed techniques based on quantum circuit splitting, the 8-qubit QCNN can be reconstructed using only 5 qubits.Experimental results demonstrate that our model achieves strong performance across 3 datasets for both binary and multiclass classification tasks. Furthermore, compared to recent technologies, our model achieves superior performance with fewer parameters, and experimental results validate the effectiveness of our model.
2501.06226
asanAI: In-Browser, No-Code, Offline-First Machine Learning Toolkit
cs.LG cs.AI cs.SE
Machine learning (ML) has become crucial in modern life, with growing interest from researchers and the public. Despite its potential, a significant entry barrier prevents widespread adoption, making it challenging for non-experts to understand and implement ML techniques. The increasing desire to leverage ML is counterbalanced by its technical complexity, creating a gap between potential and practical application. This work introduces asanAI, an offline-first, open-source, no-code machine learning toolkit designed for users of all skill levels. It allows individuals to design, debug, train, and test ML models directly in a web browser, eliminating the need for software installations and coding. The toolkit runs on any device with a modern web browser, including smartphones, and ensures user privacy through local computations while utilizing WebGL for enhanced GPU performance. Users can quickly experiment with neural networks and train custom models using various data sources, supported by intuitive visualizations of network structures and data flows. asanAI simplifies the teaching of ML concepts in educational settings and is released under an open-source MIT license, encouraging modifications. It also supports exporting models in industry-ready formats, empowering a diverse range of users to effectively learn and apply machine learning in their projects. The proposed toolkit is successfully utilized by researchers of ScaDS.AI to swiftly draft and test machine learning ideas, by trainers to effectively educate enthusiasts, and by teachers to introduce contemporary ML topics in classrooms with minimal effort and high clarity.
2501.06227
Generating and Detecting Various Types of Fake Image and Audio Content: A Review of Modern Deep Learning Technologies and Tools
cs.CR cs.LG
This paper reviews the state-of-the-art in deepfake generation and detection, focusing on modern deep learning technologies and tools based on the latest scientific advancements. The rise of deepfakes, leveraging techniques like Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), Diffusion models and other generative models, presents significant threats to privacy, security, and democracy. This fake media can deceive individuals, discredit real people and organizations, facilitate blackmail, and even threaten the integrity of legal, political, and social systems. Therefore, finding appropriate solutions to counter the potential threats posed by this technology is essential. We explore various deepfake methods, including face swapping, voice conversion, reenactment and lip synchronization, highlighting their applications in both benign and malicious contexts. The review critically examines the ongoing "arms race" between deepfake generation and detection, analyzing the challenges in identifying manipulated contents. By examining current methods and highlighting future research directions, this paper contributes to a crucial understanding of this rapidly evolving field and the urgent need for robust detection strategies to counter the misuse of this powerful technology. While focusing primarily on audio, image, and video domains, this study allows the reader to easily grasp the latest advancements in deepfake generation and detection.
2501.06229
Open-Source Manually Annotated Vocal Tract Database for Automatic Segmentation from 3D MRI Using Deep Learning: Benchmarking 2D and 3D Convolutional and Transformer Networks
cs.CV cs.SD eess.AS
Accurate segmentation of the vocal tract from magnetic resonance imaging (MRI) data is essential for various voice and speech applications. Manual segmentation is time intensive and susceptible to errors. This study aimed to evaluate the efficacy of deep learning algorithms for automatic vocal tract segmentation from 3D MRI.
2501.06230
BEN: Using Confidence-Guided Matting for Dichotomous Image Segmentation
cs.CV eess.IV
Current approaches to dichotomous image segmentation (DIS) treat image matting and object segmentation as fundamentally different tasks. As improvements in image segmentation become increasingly challenging to achieve, combining image matting and grayscale segmentation techniques offers promising new directions for architectural innovation. Inspired by the possibility of aligning these two model tasks, we propose a new architectural approach for DIS called Confidence-Guided Matting (CGM). We created the first CGM model called Background Erase Network (BEN). BEN is comprised of two components: BEN Base for initial segmentation and BEN Refiner for confidence refinement. Our approach achieves substantial improvements over current state-of-the-art methods on the DIS5K validation dataset, demonstrating that matting-based refinement can significantly enhance segmentation quality. This work opens new possibilities for cross-pollination between matting and segmentation techniques in computer vision.
2501.06231
Sustainable and Intelligent Public Facility Failure Management System Based on Large Language Models
cs.AI
This paper presents a new Large Language Model (LLM)-based Smart Device Management framework, a pioneering approach designed to address the intricate challenges of managing intelligent devices within public facilities, with a particular emphasis on applications to libraries. Our framework leverages state-of-the-art LLMs to analyze and predict device failures, thereby enhancing operational efficiency and reliability. Through prototype validation in real-world library settings, we demonstrate the framework's practical applicability and its capacity to significantly reduce budgetary constraints on public facilities. The advanced and innovative nature of our model is evident from its successful implementation in prototype testing. We plan to extend the framework's scope to include a wider array of public facilities and to integrate it with cutting-edge cybersecurity technologies, such as Internet of Things (IoT) security and machine learning algorithms for threat detection and response. This will result in a comprehensive and proactive maintenance system that not only bolsters the security of intelligent devices but also utilizes machine learning for automated analysis and real-time threat mitigation. By incorporating these advanced cybersecurity elements, our framework will be well-positioned to tackle the dynamic challenges of modern public infrastructure, ensuring robust protection against potential threats and enabling facilities to anticipate and prevent failures, leading to substantial cost savings and enhanced service quality.
2501.06232
An Interpretable ML-based Model for Predicting p-y Curves of Monopile Foundations in Sand
cs.LG cond-mat.soft
Predicting the lateral pile response is challenging due to the complexity of pile-soil interactions. Machine learning (ML) techniques have gained considerable attention for their effectiveness in non-linear analysis and prediction. This study develops an interpretable ML-based model for predicting p-y curves of monopile foundations. An XGBoost model was trained using a database compiled from existing research. The results demonstrate that the model achieves superior predictive accuracy. Shapley Additive Explanations (SHAP) was employed to enhance interpretability. The SHAP value distributions for each variable demonstrate strong alignment with established theoretical knowledge on factors affecting the lateral response of pile foundations.
2501.06233
Mechanics and Design of Metastructured Auxetic Patches with Bio-inspired Materials
cs.LG cond-mat.mtrl-sci
Metastructured auxetic patches, characterized by negative Poisson's ratios, offer unique mechanical properties that closely resemble the behavior of human tissues and organs. As a result, these patches have gained significant attention for their potential applications in organ repair and tissue regeneration. This study focuses on neural networks-based computational modeling of auxetic patches with a sinusoidal metastructure fabricated from silk fibroin, a bio-inspired material known for its biocompatibility and strength. The primary objective of this research is to introduce a novel, data-driven framework for patch design. To achieve this, we conducted experimental fabrication and mechanical testing to determine material properties and validate the corresponding finite element models. Finite element simulations were then employed to generate the necessary data, while greedy sampling, an active learning technique, was utilized to reduce the computational cost associated with data labeling. Two neural networks were trained to accurately predict Poisson's ratios and stresses for strains up to 15\%, respectively. Both models achieved $R^2$ scores exceeding 0.995, which indicates highly reliable predictions. Building on this, we developed a neural network-based design model capable of tailoring patch designs to achieve specific mechanical properties. This model demonstrated superior performance when compared to traditional optimization methods, such as genetic algorithms, by providing more efficient and precise design solutions. The proposed framework represents a significant advancement in the design of bio-inspired metastructures for medical applications, paving the way for future innovations in tissue engineering and regenerative medicine.
2501.06235
NextStop: An Improved Tracker For Panoptic LIDAR Segmentation Data
cs.CV cs.AI cs.RO
4D panoptic LiDAR segmentation is essential for scene understanding in autonomous driving and robotics ,combining semantic and instance segmentation with temporal consistency.Current methods, like 4D-PLS and 4D-STOP, use a tracking-by-detection methodology, employing deep learning networks to perform semantic and instance segmentation on each frame. To maintain temporal consistency, large-size instances detected in the current frame are compared and associated with instances within a temporal window that includes the current and preceding frames. However, their reliance on short-term instance detection, lack of motion estimation, and exclusion of small-sized instances lead to frequent identity switches and reduced tracking performance. We address these issues with the NextStop1 tracker, which integrates Kalman filter-based motion estimation, data association, and lifespan management, along with a tracklet state concept to improve prioritization. Evaluated using the LiDAR Segmentation and Tracking Quality (LSTQ) metric on the SemanticKITTI validation set, NextStop demonstrated enhanced tracking performance, particularly for small-sized objects like people and bicyclists, with fewer ID switches, earlier tracking initiation, and improved reliability in complex environments. The source code is available at https://github.com/AIROTAU/NextStopTracker
2501.06236
Data-Driven Radio Propagation Modeling using Graph Neural Networks
cs.LG cs.AI cs.NI
Modeling radio propagation is essential for wireless network design and performance optimization. Traditional methods rely on physics models of radio propagation, which can be inaccurate or inflexible. In this work, we propose using graph neural networks to learn radio propagation behaviors directly from real-world network data. Our approach converts the radio propagation environment into a graph representation, with nodes corresponding to locations and edges representing spatial and ray-tracing relationships between locations. The graph is generated by converting images of the environment into a graph structure, with specific relationships between nodes. The model is trained on this graph representation, using sensor measurements as target data. We demonstrate that the graph neural network, which learns to predict radio propagation directly from data, achieves competitive performance compared to traditional heuristic models. This data-driven approach outperforms classic numerical solvers in terms of both speed and accuracy. To the best of our knowledge, we are the first to apply graph neural networks to real-world radio propagation data to generate coverage maps, enabling generative models of signal propagation with point measurements only.
2501.06237
Forecasting Anonymized Electricity Load Profiles
cs.CR cs.AI cs.LG
In the evolving landscape of data privacy, the anonymization of electric load profiles has become a critical issue, especially with the enforcement of the General Data Protection Regulation (GDPR) in Europe. These electric load profiles, which are essential datasets in the energy industry, are classified as personal behavioral data, necessitating stringent protective measures. This article explores the implications of this classification, the importance of data anonymization, and the potential of forecasting using microaggregated data. The findings underscore that effective anonymization techniques, such as microaggregation, do not compromise the performance of forecasting models under certain conditions (i.e., forecasting aggregated). In such an aggregated level, microaggregated data maintains high levels of utility, with minimal impact on forecasting accuracy. The implications for the energy sector are profound, suggesting that privacy-preserving data practices can be integrated into smart metering technology applications without hindering their effectiveness.
2501.06238
Multi-field Visualization: Trait design and trait-induced merge trees
cs.LG cs.GR
Feature level sets (FLS) have shown significant potential in the analysis of multi-field data by using traits defined in attribute space to specify features in the domain. In this work, we address key challenges in the practical use of FLS: trait design and feature selection for rendering. To simplify trait design, we propose a Cartesian decomposition of traits into simpler components, making the process more intuitive and computationally efficient. Additionally, we utilize dictionary learning results to automatically suggest point traits. To enhance feature selection, we introduce trait-induced merge trees (TIMTs), a generalization of merge trees for feature level sets, aimed at topologically analyzing tensor fields or general multi-variate data. The leaves in the TIMT represent areas in the input data that are closest to the defined trait, thereby most closely resembling the defined feature. This merge tree provides a hierarchy of features, enabling the querying of the most relevant and persistent features. Our method includes various query techniques for the tree, allowing the highlighting of different aspects. We demonstrate the cross-application capabilities of this approach through five case studies from different domains.
2501.06239
Towards a scalable AI-driven framework for data-independent Cyber Threat Intelligence Information Extraction
cs.CR cs.AI cs.CL
Cyber Threat Intelligence (CTI) is critical for mitigating threats to organizations, governments, and institutions, yet the necessary data are often dispersed across diverse formats. AI-driven solutions for CTI Information Extraction (IE) typically depend on high-quality, annotated data, which are not always available. This paper introduces 0-CTI, a scalable AI-based framework designed for efficient CTI Information Extraction. Leveraging advanced Natural Language Processing (NLP) techniques, particularly Transformer-based architectures, the proposed system processes complete text sequences of CTI reports to extract a cyber ontology of named entities and their relationships. Our contribution is the development of 0-CTI, the first modular framework for CTI Information Extraction that supports both supervised and zero-shot learning. Unlike existing state-of-the-art models that rely heavily on annotated datasets, our system enables fully dataless operation through zero-shot methods for both Entity and Relation Extraction, making it adaptable to various data availability scenarios. Additionally, our supervised Entity Extractor surpasses current state-of-the-art performance in cyber Entity Extraction, highlighting the dual strength of the framework in both low-resource and data-rich environments. By aligning the system's outputs with the Structured Threat Information Expression (STIX) format, a standard for information exchange in the cybersecurity domain, 0-CTI standardizes extracted knowledge, enhancing communication and collaboration in cybersecurity operations.