id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2501.06478
Speech Recognition for Automatically Assessing Afrikaans and isiXhosa Preschool Oral Narratives
eess.AS cs.CL cs.SD
We develop automatic speech recognition (ASR) systems for stories told by Afrikaans and isiXhosa preschool children. Oral narratives provide a way to assess children's language development before they learn to read. We consider a range of prior child-speech ASR strategies to determine which is best suited to this unique setting. Using Whisper and only 5 minutes of transcribed in-domain child speech, we find that additional in-domain adult data (adult speech matching the story domain) provides the biggest improvement, especially when coupled with voice conversion. Semi-supervised learning also helps for both languages, while parameter-efficient fine-tuning helps on Afrikaans but not on isiXhosa (which is under-represented in the Whisper model). Few child-speech studies look at non-English data, and even fewer at the preschool ages of 4 and 5. Our work therefore represents a unique validation of a wide range of previous child-speech ASR strategies in an under-explored setting.
2501.06480
Flash Window Attention: speedup the attention computation for Swin Transformer
cs.CV
To address the high resolution of image pixels, the Swin Transformer introduces window attention. This mechanism divides an image into non-overlapping windows and restricts attention computation to within each window, significantly enhancing computational efficiency. To further optimize this process, one might consider replacing standard attention with flash attention, which has proven to be more efficient in language models. However, a direct substitution is ineffective. Flash attention is designed for long sequences, whereas window attention deals with shorter sequences but must handle numerous of them in parallel. In this report, we present an optimized solution called Flash Window Attention, tailored specifically for window attention. Flash Window Attention improves attention computation efficiency by up to 300% and enhances end-to-end runtime efficiency by up to 30%. Our code is available online.
2501.06481
Focus-N-Fix: Region-Aware Fine-Tuning for Text-to-Image Generation
cs.CV
Text-to-image (T2I) generation has made significant advances in recent years, but challenges still remain in the generation of perceptual artifacts, misalignment with complex prompts, and safety. The prevailing approach to address these issues involves collecting human feedback on generated images, training reward models to estimate human feedback, and then fine-tuning T2I models based on the reward models to align them with human preferences. However, while existing reward fine-tuning methods can produce images with higher rewards, they may change model behavior in unexpected ways. For example, fine-tuning for one quality aspect (e.g., safety) may degrade other aspects (e.g., prompt alignment), or may lead to reward hacking (e.g., finding a way to increase rewards without having the intended effect). In this paper, we propose Focus-N-Fix, a region-aware fine-tuning method that trains models to correct only previously problematic image regions. The resulting fine-tuned model generates images with the same high-level structure as the original model but shows significant improvements in regions where the original model was deficient in safety (over-sexualization and violence), plausibility, or other criteria. Our experiments demonstrate that Focus-N-Fix improves these localized quality aspects with little or no degradation to others and typically imperceptible changes in the rest of the image. Disclaimer: This paper contains images that may be overly sexual, violent, offensive, or harmful.
2501.06485
A Diffusive Data Augmentation Framework for Reconstruction of Complex Network Evolutionary History
cs.AI
The evolutionary processes of complex systems contain critical information regarding their functional characteristics. The generation time of edges provides insights into the historical evolution of various networked complex systems, such as protein-protein interaction networks, ecosystems, and social networks. Recovering these evolutionary processes holds significant scientific value, including aiding in the interpretation of the evolution of protein-protein interaction networks. However, existing methods are capable of predicting the generation times of remaining edges given a partial temporal network but often perform poorly in cross-network prediction tasks. These methods frequently fail in edge generation time recovery tasks for static networks that lack timestamps. In this work, we adopt a comparative paradigm-based framework that fuses multiple networks for training, enabling cross-network learning of the relationship between network structure and edge generation times. Compared to separate training, this approach yields an average accuracy improvement of 16.98%. Furthermore, given the difficulty in collecting temporal networks, we propose a novel diffusion-model-based generation method to produce a large number of temporal networks. By combining real temporal networks with generated ones for training, we achieve an additional average accuracy improvement of 5.46% through joint training.
2501.06488
NVS-SQA: Exploring Self-Supervised Quality Representation Learning for Neurally Synthesized Scenes without References
cs.CV cs.AI cs.HC cs.MM eess.IV
Neural View Synthesis (NVS), such as NeRF and 3D Gaussian Splatting, effectively creates photorealistic scenes from sparse viewpoints, typically evaluated by quality assessment methods like PSNR, SSIM, and LPIPS. However, these full-reference methods, which compare synthesized views to reference views, may not fully capture the perceptual quality of neurally synthesized scenes (NSS), particularly due to the limited availability of dense reference views. Furthermore, the challenges in acquiring human perceptual labels hinder the creation of extensive labeled datasets, risking model overfitting and reduced generalizability. To address these issues, we propose NVS-SQA, a NSS quality assessment method to learn no-reference quality representations through self-supervision without reliance on human labels. Traditional self-supervised learning predominantly relies on the "same instance, similar representation" assumption and extensive datasets. However, given that these conditions do not apply in NSS quality assessment, we employ heuristic cues and quality scores as learning objectives, along with a specialized contrastive pair preparation process to improve the effectiveness and efficiency of learning. The results show that NVS-SQA outperforms 17 no-reference methods by a large margin (i.e., on average 109.5% in SRCC, 98.6% in PLCC, and 91.5% in KRCC over the second best) and even exceeds 16 full-reference methods across all evaluation metrics (i.e., 22.9% in SRCC, 19.1% in PLCC, and 18.6% in KRCC over the second best).
2501.06490
Sequential Classification of Aviation Safety Occurrences with Natural Language Processing
cs.CL cs.LG
Safety is a critical aspect of the air transport system given even slight operational anomalies can result in serious consequences. To reduce the chances of aviation safety occurrences, accidents and incidents are reported to establish the root cause, propose safety recommendations etc. However, analysis narratives of the pre-accident events are presented using human-understandable, raw, unstructured, text that a computer system cannot understand. The ability to classify and categorise safety occurrences from their textual narratives would help aviation industry stakeholders make informed safety-critical decisions. To classify and categorise safety occurrences, we applied natural language processing (NLP) and AI (Artificial Intelligence) models to process text narratives. The study aimed to answer the question. How well can the damage level caused to the aircraft in a safety occurrence be inferred from the text narrative using natural language processing. The classification performance of various deep learning models including LSTM, BLSTM, GRU, sRNN, and combinations of these models including LSTM and GRU, BLSTM+GRU, sRNN and LSTM, sRNN and BLSTM, sRNN and GRU, sRNN and BLSTM and GRU, and sRNN and LSTM and GRU was evaluated on a set of 27,000 safety occurrence reports from the NTSB. The results of this study indicate that all models investigated performed competitively well recording an accuracy of over 87.9% which is well above the random guess of 25% for a four-class classification problem. Also, the models recorded high precision, recall, and F1 scores above 80%, 88%, and 85%, respectively. sRNN slightly outperformed other single models in terms of recall (90%) and accuracy (90%) while LSTM reported slightly better performance in terms of precision (87%).
2501.06491
Improving Requirements Classification with SMOTE-Tomek Preprocessing
cs.SE cs.AI cs.SY eess.SY
This study emphasizes the domain of requirements engineering by applying the SMOTE-Tomek preprocessing technique, combined with stratified K-fold cross-validation, to address class imbalance in the PROMISE dataset. This dataset comprises 969 categorized requirements, classified into functional and non-functional types. The proposed approach enhances the representation of minority classes while maintaining the integrity of validation folds, leading to a notable improvement in classification accuracy. Logistic regression achieved 76.16\%, significantly surpassing the baseline of 58.31\%. These results highlight the applicability and efficiency of machine learning models as scalable and interpretable solutions.
2501.06492
A New Flexible Train-Test Split Algorithm, an approach for choosing among the Hold-out, K-fold cross-validation, and Hold-out iteration
cs.LG
Artificial Intelligent transformed industries, like engineering, medicine, finance. Predictive models use supervised learning, a vital Machine learning subset. Crucial for model evaluation, cross-validation includes re-substitution, hold-out, and K-fold. This study focuses on improving the accuracy of ML algorithms across three different datasets. To evaluate Hold-out, Hold-out with iteration, and K-fold Cross-Validation techniques, we created a flexible Python program. By modifying parameters like test size, Random State, and 'k' values, we were able to improve accuracy assessment. The outcomes demonstrate the Hold-out validation method's persistent superiority, particularly with a test size of 10%. With iterations and Random State settings, hold-out with iteration shows little accuracy variance. It suggests that there are variances according to algorithm, with Decision Tree doing best for Framingham and Naive Bayes and K Nearest Neighbors for COVID-19. Different datasets require different optimal K values in K-Fold Cross Validation, highlighting these considerations. This study challenges the universality of K values in K-Fold Cross Validation and suggests a 10% test size and 90% training size for better outcomes. It also emphasizes the contextual impact of dataset features, sample size, feature count, and selected methodologies. Researchers can adapt these codes for their dataset to obtain highest accuracy with specific evaluation.
2501.06493
Whole-Body Integrated Motion Planning for Aerial Manipulators
cs.RO
Efficient motion planning for Aerial Manipulators (AMs) is essential for tackling complex manipulation tasks, yet achieving coupled trajectory planning remains challenging. In this work, we propose, to the best of our knowledge, the first whole-body integrated motion planning framework for aerial manipulators, which is facilitated by an improved Safe Flight Corridor (SFC) generation strategy and high-dimensional collision-free trajectory planning. In particular, we formulate an optimization problem to generate feasible trajectories for both the quadrotor and manipulator while ensuring collision avoidance, dynamic feasibility, kinematic feasibility, and waypoint constraints. To achieve collision avoidance, we introduce a variable geometry approximation method, which dynamically models the changing collision volume induced by different manipulator configurations. Moreover, waypoint constraints in our framework are defined in $\mathrm{SE(3)\times\mathbb{R}^3}$, allowing the aerial manipulator to traverse specified positions while maintaining desired attitudes and end-effector states. The effectiveness of our framework is validated through comprehensive simulations and real-world experiments across various environments.
2501.06494
TopoFormer: Integrating Transformers and ConvLSTMs for Coastal Topography Prediction
eess.SP cs.AI
This paper presents \textit{TopoFormer}, a novel hybrid deep learning architecture that integrates transformer-based encoders with convolutional long short-term memory (ConvLSTM) layers for the precise prediction of topographic beach profiles referenced to elevation datums, with a particular focus on Mean Low Water Springs (MLWS) and Mean Low Water Neaps (MLWN). Accurate topographic estimation down to MLWS is critical for coastal management, navigation safety, and environmental monitoring. Leveraging a comprehensive dataset from the Wales Coastal Monitoring Centre (WCMC), consisting of over 2000 surveys across 36 coastal survey units, TopoFormer addresses key challenges in topographic prediction, including temporal variability and data gaps in survey measurements. The architecture uniquely combines multi-head attention mechanisms and ConvLSTM layers to capture both long-range dependencies and localized temporal patterns inherent in beach profiles data. TopoFormer's predictive performance was rigorously evaluated against state-of-the-art models, including DenseNet, 1D/2D CNNs, and LSTMs. While all models demonstrated strong performance, \textit{TopoFormer} achieved the lowest mean absolute error (MAE), as low as 2 cm, and provided superior accuracy in both in-distribution (ID) and out-of-distribution (OOD) evaluations.
2501.06496
Analyzing the Role of Context in Forecasting with Large Language Models
cs.CL cs.IR
This study evaluates the forecasting performance of recent language models (LLMs) on binary forecasting questions. We first introduce a novel dataset of over 600 binary forecasting questions, augmented with related news articles and their concise question-related summaries. We then explore the impact of input prompts with varying level of context on forecasting performance. The results indicate that incorporating news articles significantly improves performance, while using few-shot examples leads to a decline in accuracy. We find that larger models consistently outperform smaller models, highlighting the potential of LLMs in enhancing automated forecasting.
2501.06497
PASS: Presentation Automation for Slide Generation and Speech
cs.CL cs.AI
In today's fast-paced world, effective presentations have become an essential tool for communication in both online and offline meetings. The crafting of a compelling presentation requires significant time and effort, from gathering key insights to designing slides that convey information clearly and concisely. However, despite the wealth of resources available, people often find themselves manually extracting crucial points, analyzing data, and organizing content in a way that ensures clarity and impact. Furthermore, a successful presentation goes beyond just the slides; it demands rehearsal and the ability to weave a captivating narrative to fully engage the audience. Although there has been some exploration of automating document-to-slide generation, existing research is largely centered on converting research papers. In addition, automation of the delivery of these presentations has yet to be addressed. We introduce PASS, a pipeline used to generate slides from general Word documents, going beyond just research papers, which also automates the oral delivery of the generated slides. PASS analyzes user documents to create a dynamic, engaging presentation with an AI-generated voice. Additionally, we developed an LLM-based evaluation metric to assess our pipeline across three critical dimensions of presentations: relevance, coherence, and redundancy. The data and codes are available at https://github.com/AggarwalTushar/PASS.
2501.06505
Online Algorithm for Aggregating Experts' Predictions with Unbounded Quadratic Loss
cs.LG
We consider the problem of online aggregation of expert predictions with the quadratic loss function. We propose an algorithm for aggregating expert predictions which does not require a prior knowledge of the upper bound on the losses. The algorithm is based on the exponential reweighing of expert losses.
2501.06506
Resource Allocation under the Latin Square Constraint
cs.GT cs.AI cs.MA
A Latin square is an $n \times n$ matrix filled with $n$ distinct symbols, each of which appears exactly once in each row and exactly once in each column. We introduce a problem of allocating $n$ indivisible items among $n$ agents over $n$ rounds while satisfying the Latin square constraint. This constraint ensures that each agent receives no more than one item per round and receives each item at most once. Each agent has an additive valuation on the item--round pairs. Real-world applications like scheduling, resource management, and experimental design require the Latin square constraint to satisfy fairness or balancedness in allocation. Our goal is to find a partial or complete allocation that maximizes the sum of the agents' valuations (utilitarian social welfare) or the minimum of the agents' valuations (egalitarian social welfare). For the problem of maximizing utilitarian social welfare, we prove NP-hardness even when the valuations are binary additive. We then provide $(1-1/e)$ and $(1-1/e)/4$-approximation algorithms for partial and complete settings, respectively. Additionally, we present fixed-parameter tractable (FPT) algorithms with respect to the order of Latin square and the optimum value for both partial and complete settings. For the problem of maximizing egalitarian social welfare, we establish that deciding whether the optimum value is at most $1$ or at least $2$ is NP-hard for both the partial and complete settings, even when the valuations are binary. Furthermore, we demonstrate that checking the existence of a complete allocation that satisfies each of envy-free, proportional, equitable, envy-free up to any good, proportional up to any good, or equitable up to any good is NP-hard, even when the valuations are identical.
2501.06510
Cooperative Optimal Output Tracking for Discrete-Time Multiagent Systems: Stabilizing Policy Iteration Frameworks and Analysis
eess.SY cs.SY
In this paper, two model-free optimal output tracking frameworks based on policy iteration for discrete-time multi-agent systems are proposed. First, we establish a framework of stabilizing policy iteration that can start from any initial feedback control policy, relaxing the dependence of traditional policy iteration on the initial stabilizing control policy. Then, another efficient and equivalent $Q$-learning policy iteration framework is developed, which is shown to require only less system data to get the same results as the stabilizing policy iteration. Both frameworks obtain stabilizing control policy by iterating the stabilizing virtual closed-loop system step-by-step to the actual closed-loop system. Multiple explicit schemes for the iteration step-size/coefficient are designed and their stability during the above iterations is analyzed. By using the generated closed-loop stabilizing control policy and two frameworks, the optimal feedback control gain is obtained. The approximate solution of the regulator equations is found by model-free iteration, which leads to the optimal feedforward gain. Finally, the cooperative optimal output tracking is realized by a distributed feedforward-feedback controller. The proposed algorithms are validated by simulation.
2501.06514
Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition
cs.SD cs.AI eess.AS
Current research in audio deepfake detection is gradually transitioning from binary classification to multi-class tasks, referred as audio deepfake source tracing task. However, existing studies on source tracing consider only closed-set scenarios and have not considered the challenges posed by open-set conditions. In this paper, we define the Neural Codec Source Tracing (NCST) task, which is capable of performing open-set neural codec classification and interpretable ALM detection. Specifically, we constructed the ST-Codecfake dataset for the NCST task, which includes bilingual audio samples generated by 11 state-of-the-art neural codec methods and ALM-based out-ofdistribution (OOD) test samples. Furthermore, we establish a comprehensive source tracing benchmark to assess NCST models in open-set conditions. The experimental results reveal that although the NCST models perform well in in-distribution (ID) classification and OOD detection, they lack robustness in classifying unseen real audio. The ST-codecfake dataset and code are available.
2501.06521
Fine-tuning Large Language Models for Improving Factuality in Legal Question Answering
cs.CL
Hallucination, or the generation of incorrect or fabricated information, remains a critical challenge in large language models (LLMs), particularly in high-stake domains such as legal question answering (QA). In order to mitigate the hallucination rate in legal QA, we first introduce a benchmark called LegalHalBench and three automatic metrics to evaluate the common hallucinations when LLMs answer legal questions. We then propose a hallucination mitigation method that integrates behavior cloning and a novel Hard Sample-aware Iterative Direct Preference Optimization (HIPO). We conduct extensive real-data experiments to validate the effectiveness of our approach. Our results demonstrate remarkable improvements in various metrics, including the newly proposed Non-Hallucinated Statute Rate, Statute Relevance Rate, Legal Claim Truthfulness, as well as traditional metrics such as METEOR, BERTScore, ROUGE-L, and win rates.
2501.06524
Multi-View Factorizing and Disentangling: A Novel Framework for Incomplete Multi-View Multi-Label Classification
cs.CV
Multi-view multi-label classification (MvMLC) has recently garnered significant research attention due to its wide range of real-world applications. However, incompleteness in views and labels is a common challenge, often resulting from data collection oversights and uncertainties in manual annotation. Furthermore, the task of learning robust multi-view representations that are both view-consistent and view-specific from diverse views still a challenge problem in MvMLC. To address these issues, we propose a novel framework for incomplete multi-view multi-label classification (iMvMLC). Our method factorizes multi-view representations into two independent sets of factors: view-consistent and view-specific, and we correspondingly design a graph disentangling loss to fully reduce redundancy between these representations. Additionally, our framework innovatively decomposes consistent representation learning into three key sub-objectives: (i) how to extract view-shared information across different views, (ii) how to eliminate intra-view redundancy in consistent representations, and (iii) how to preserve task-relevant information. To this end, we design a robust task-relevant consistency learning module that collaboratively learns high-quality consistent representations, leveraging a masked cross-view prediction (MCP) strategy and information theory. Notably, all modules in our framework are developed to function effectively under conditions of incomplete views and labels, making our method adaptable to various multi-view and multi-label datasets. Extensive experiments on five datasets demonstrate that our method outperforms other leading approaches.
2501.06527
Scaffolding Creativity: Integrating Generative AI Tools and Real-world Experiences in Business Education
cs.AI cs.HC
This case study explores the integration of Generative AI tools and real-world experiences in business education. Through a study of an innovative undergraduate course, we investigate how AI-assisted learning, combined with experiential components, impacts students' creative processes and learning outcomes. Our findings reveal that this integrated approach accelerates knowledge acquisition, enables students to overcome traditional creative barriers, and facilitates a dynamic interplay between AI-generated insights and real-world observations. The study also highlights challenges, including the need for instructors with high AI literacy and the rapid evolution of AI tools creating a moving target for curriculum design. These insights contribute to the growing body of literature on AI in education and provide actionable recommendations for educators preparing students for the complexities of modern business environments.
2501.06528
Safe Circumnavigation of a Hostile Target Using Range-Based Measurements
cs.RO cs.SY eess.SY
Robotic systems are frequently deployed in missions that are dull, dirty, and dangerous, where ensuring their safety is of paramount importance when designing stabilizing controllers to achieve their desired goals. This paper addresses the problem of safe circumnavigation around a hostile target by a nonholonomic robot, with the objective of maintaining a desired safe distance from the target. Our solution approach involves incorporating an auxiliary circle into the problem formulation, which assists in navigating the robot around the target using available range-based measurements. By leveraging the concept of a barrier Lyapunov function, we propose a novel control law that ensures stable circumnavigation around the target while preventing the robot from entering the safety circle. This controller is designed based on a parameter that depends on the radii of three circles, namely the stabilizing circle, the auxiliary circle, and the safety circle. By identifying an appropriate range for this design parameter, we rigorously prove the stability of the desired equilibrium of the closed-loop system. Additionally, we provide an analysis of the robot's motion within the auxiliary circle, which is influenced by a gain parameter in the proposed controller. Simulation and experimental results are presented to illustrate the key theoretical developments.
2501.06532
Determination of galaxy photometric redshifts using Conditional Generative Adversarial Networks (CGANs)
astro-ph.IM astro-ph.CO cs.AI
Accurate and reliable photometric redshifts determination is one of the key aspects for wide-field photometric surveys. Determination of photometric redshift for galaxies, has been traditionally solved by use of machine-learning and artificial intelligence techniques trained on a calibration sample of galaxies, where both photometry and spectrometry are determined. On this paper, we present a new algorithmic approach for determining photometric redshifts of galaxies using Conditional Generative Adversarial Networks (CGANs). Proposed CGAN implementation, approaches photometric redshift determination as a probabilistic regression, where instead of determining a single value for the estimated redshift of the galaxy, a full probability density is computed. The methodology proposed, is tested with data from Dark Energy Survey (DES) Y1 data and compared with other existing algorithm such as a Random Forest regressor.
2501.06533
DivTrackee versus DynTracker: Promoting Diversity in Anti-Facial Recognition against Dynamic FR Strategy
cs.CV cs.CR
The widespread adoption of facial recognition (FR) models raises serious concerns about their potential misuse, motivating the development of anti-facial recognition (AFR) to protect user facial privacy. In this paper, we argue that the static FR strategy, predominantly adopted in prior literature for evaluating AFR efficacy, cannot faithfully characterize the actual capabilities of determined trackers who aim to track a specific target identity. In particular, we introduce \emph{\ourAttack}, a dynamic FR strategy where the model's gallery database is iteratively updated with newly recognized target identity images. Surprisingly, such a simple approach renders all the existing AFR protections ineffective. To mitigate the privacy threats posed by DynTracker, we advocate for explicitly promoting diversity in the AFR-protected images. We hypothesize that the lack of diversity is the primary cause of the failure of existing AFR methods. Specifically, we develop \emph{DivTrackee}, a novel method for crafting diverse AFR protections that builds upon a text-guided image generation framework and diversity-promoting adversarial losses. Through comprehensive experiments on various facial image benchmarks and feature extractors, we demonstrate DynTracker's strength in breaking existing AFR methods and the superiority of DivTrackee in preventing user facial images from being identified by dynamic FR strategies. We believe our work can act as an important initial step towards developing more effective AFR methods for protecting user facial privacy against determined trackers.
2501.06534
Dynamic Causal Structure Discovery and Causal Effect Estimation
stat.ML cs.LG
To represent the causal relationships between variables, a directed acyclic graph (DAG) is widely utilized in many areas, such as social sciences, epidemics, and genetics. Many causal structure learning approaches are developed to learn the hidden causal structure utilizing deep-learning approaches. However, these approaches have a hidden assumption that the causal relationship remains unchanged over time, which may not hold in real life. In this paper, we develop a new framework to model the dynamic causal graph where the causal relations are allowed to be time-varying. We incorporate the basis approximation method into the score-based causal discovery approach to capture the dynamic pattern of the causal graphs. Utilizing the autoregressive model structure, we could capture both contemporaneous and time-lagged causal relationships while allowing them to vary with time. We propose an algorithm that could provide both past-time estimates and future-time predictions on the causal graphs, and conduct simulations to demonstrate the usefulness of the proposed method. We also apply the proposed method for the covid-data analysis, and provide causal estimates on how policy restriction's effect changes.
2501.06536
Dispersion Measures as Predictors of Lexical Decision Time, Word Familiarity, and Lexical Complexity
cs.CL
Various measures of dispersion have been proposed to paint a fuller picture of a word's distribution in a corpus, but only little has been done to validate them externally. We evaluate a wide range of dispersion measures as predictors of lexical decision time, word familiarity, and lexical complexity in five diverse languages. We find that the logarithm of range is not only a better predictor than log-frequency across all tasks and languages, but that it is also the most powerful additional variable to log-frequency, consistently outperforming the more complex dispersion measures. We discuss the effects of corpus part granularity and logarithmic transformation, shedding light on contradictory results of previous studies.
2501.06540
CeViT: Copula-Enhanced Vision Transformer in multi-task learning and bi-group image covariates with an application to myopia screening
cs.CV math.ST stat.AP stat.ME stat.TH
We aim to assist image-based myopia screening by resolving two longstanding problems, "how to integrate the information of ocular images of a pair of eyes" and "how to incorporate the inherent dependence among high-myopia status and axial length for both eyes." The classification-regression task is modeled as a novel 4-dimensional muti-response regression, where discrete responses are allowed, that relates to two dependent 3rd-order tensors (3D ultrawide-field fundus images). We present a Vision Transformer-based bi-channel architecture, named CeViT, where the common features of a pair of eyes are extracted via a shared Transformer encoder, and the interocular asymmetries are modeled through separated multilayer perceptron heads. Statistically, we model the conditional dependence among mixture of discrete-continuous responses given the image covariates by a so-called copula loss. We establish a new theoretical framework regarding fine-tuning on CeViT based on latent representations, allowing the black-box fine-tuning procedure interpretable and guaranteeing higher relative efficiency of fine-tuning weight estimation in the asymptotic setting. We apply CeViT to an annotated ultrawide-field fundus image dataset collected by Shanghai Eye \& ENT Hospital, demonstrating that CeViT enhances the baseline model in both accuracy of classifying high-myopia and prediction of AL on both eyes.
2501.06545
Energy-Aware Resource Allocation for Energy Harvesting Powered Wireless Sensor Nodes
cs.IT eess.SP math.IT
Low harvested energy poses a significant challenge to sustaining continuous communication in energy harvesting (EH)-powered wireless sensor networks. This is mainly due to intermittent and limited power availability from radio frequency signals. In this paper, we introduce a novel energy-aware resource allocation problem aimed at enabling the asynchronous accumulate-then-transmit protocol, offering an alternative to the extensively studied harvest-then-transmit approach. Specifically, we jointly optimize power allocation and time fraction dedicated to EH to maximize the average long-term system throughput, accounting for both data and energy queue lengths. By leveraging inner approximation and network utility maximization techniques, we develop a simple yet efficient iterative algorithm that guarantees at least a local optimum and achieves long-term utility improvement. Numerical results highlight the proposed approach's effectiveness in terms of both queue length and sustained system throughput.
2501.06546
Natural Language Supervision for Low-light Image Enhancement
cs.CV cs.AI
With the development of deep learning, numerous methods for low-light image enhancement (LLIE) have demonstrated remarkable performance. Mainstream LLIE methods typically learn an end-to-end mapping based on pairs of low-light and normal-light images. However, normal-light images under varying illumination conditions serve as reference images, making it difficult to define a ``perfect'' reference image This leads to the challenge of reconciling metric-oriented and visual-friendly results. Recently, many cross-modal studies have found that side information from other related modalities can guide visual representation learning. Based on this, we introduce a Natural Language Supervision (NLS) strategy, which learns feature maps from text corresponding to images, offering a general and flexible interface for describing an image under different illumination. However, image distributions conditioned on textual descriptions are highly multimodal, which makes training difficult. To address this issue, we design a Textual Guidance Conditioning Mechanism (TCM) that incorporates the connections between image regions and sentence words, enhancing the ability to capture fine-grained cross-modal cues for images and text. This strategy not only utilizes a wider range of supervised sources, but also provides a new paradigm for LLIE based on visual and textual feature alignment. In order to effectively identify and merge features from various levels of image and textual information, we design an Information Fusion Attention (IFA) module to enhance different regions at different levels. We integrate the proposed TCM and IFA into a Natural Language Supervision network for LLIE, named NaLSuper. Finally, extensive experiments demonstrate the robustness and superior effectiveness of our proposed NaLSuper.
2501.06550
CoreNet: Conflict Resolution Network for Point-Pixel Misalignment and Sub-Task Suppression of 3D LiDAR-Camera Object Detection
cs.CV
Fusing multi-modality inputs from different sensors is an effective way to improve the performance of 3D object detection. However, current methods overlook two important conflicts: point-pixel misalignment and sub-task suppression. The former means a pixel feature from the opaque object is projected to multiple point features of the same ray in the world space, and the latter means the classification prediction and bounding box regression may cause mutual suppression. In this paper, we propose a novel method named Conflict Resolution Network (CoreNet) to address the aforementioned issues. Specifically, we first propose a dual-stream transformation module to tackle point-pixel misalignment. It consists of ray-based and point-based 2D-to-BEV transformations. Both of them achieve approximately unique mapping from the image space to the world space. Moreover, we introduce a task-specific predictor to tackle sub-task suppression. It uses the dual-branch structure which adopts class-specific query and Bbox-specific query to corresponding sub-tasks. Each task-specific query is constructed of task-specific feature and general feature, which allows the heads to adaptively select information of interest based on different sub-tasks. Experiments on the large-scale nuScenes dataset demonstrate the superiority of our proposed CoreNet, by achieving 75.6\% NDS and 73.3\% mAP on the nuScenes test set without test-time augmentation and model ensemble techniques. The ample ablation study also demonstrates the effectiveness of each component. The code is released on https://github.com/liyih/CoreNet.
2501.06552
When xURLLC Meets NOMA: A Stochastic Network Calculus Perspective
eess.SP cs.IT cs.SY eess.SY math.IT
The advent of next-generation ultra-reliable and low-latency communications (xURLLC) presents stringent and unprecedented requirements for key performance indicators (KPIs). As a disruptive technology, non-orthogonal multiple access (NOMA) harbors the potential to fulfill these stringent KPIs essential for xURLLC. However, the immaturity of research on the tail distributions of these KPIs significantly impedes the application of NOMA to xURLLC. Stochastic network calculus (SNC), as a potent methodology, is leveraged to provide dependable theoretical insights into tail distribution analysis and statistical QoS provisioning (SQP). In this article, we develop a NOMA-assisted uplink xURLLC network architecture that incorporates an SNC-based SQP theoretical framework (SNC-SQP) to support tail distribution analysis in terms of delay, age-of-information (AoI), and reliability. Based on SNC-SQP, an SQP-driven power optimization problem is proposed to minimize transmit power while guaranteeing xURLLC's KPIs on delay, AoI, reliability, and power consumption. Extensive simulations validate our proposed theoretical framework and demonstrate that the proposed power allocation scheme significantly reduces uplink transmit power and outperforms conventional schemes in terms of SQP performance.
2501.06553
VASparse: Towards Efficient Visual Hallucination Mitigation for Large Vision-Language Model via Visual-Aware Sparsification
cs.CV
Large Vision-Language Models (LVLMs) may produce outputs that are unfaithful to reality, also known as visual hallucinations (VH), which significantly impedes their real-world usage. To alleviate VH, various decoding strategies have been proposed to enhance visual information. However, many of these methods may require secondary decoding and rollback, which significantly reduces inference speed. In this work, we propose an efficient plug-and-play decoding algorithm via Visual-Aware Sparsification (VASparse) from the perspective of token sparsity for mitigating VH. VASparse is inspired by empirical observations: (1) the sparse activation of attention in LVLMs, and (2) visual-agnostic tokens sparsification exacerbates VH. Based on these insights, we propose a novel token sparsification strategy that balances efficiency and trustworthiness. Specifically, VASparse implements a visual-aware token selection strategy during decoding to reduce redundant tokens while preserving visual context effectively. Additionally, we innovatively introduce a sparse-based visual contrastive decoding method to recalibrate the distribution of hallucinated outputs without the time overhead associated with secondary decoding. Subsequently, VASparse recalibrates attention scores to penalize attention sinking of LVLMs towards text tokens. Extensive experiments across four popular benchmarks confirm the effectiveness of VASparse in mitigating VH across different LVLM families without requiring additional training or post-processing. Impressively, VASparse achieves state-of-the-art performance for mitigating VH while maintaining competitive decoding speed. Code is available at https://github.com/mengchuang123/VASparse-github.
2501.06554
Hierarchical Reinforcement Learning for Optimal Agent Grouping in Cooperative Systems
cs.LG cs.AI cs.MA
This paper presents a hierarchical reinforcement learning (RL) approach to address the agent grouping or pairing problem in cooperative multi-agent systems. The goal is to simultaneously learn the optimal grouping and agent policy. By employing a hierarchical RL framework, we distinguish between high-level decisions of grouping and low-level agents' actions. Our approach utilizes the CTDE (Centralized Training with Decentralized Execution) paradigm, ensuring efficient learning and scalable execution. We incorporate permutation-invariant neural networks to handle the homogeneity and cooperation among agents, enabling effective coordination. The option-critic algorithm is adapted to manage the hierarchical decision-making process, allowing for dynamic and optimal policy adjustments.
2501.06557
A Survey on Spoken Italian Datasets and Corpora
cs.CL cs.AI cs.DL
Spoken language datasets are vital for advancing linguistic research, Natural Language Processing, and speech technology. However, resources dedicated to Italian, a linguistically rich and diverse Romance language, remain underexplored compared to major languages like English or Mandarin. This survey provides a comprehensive analysis of 66 spoken Italian datasets, highlighting their characteristics, methodologies, and applications. The datasets are categorized by speech type, source and context, and demographic and linguistic features, with a focus on their utility in fields such as Automatic Speech Recognition, emotion detection, and education. Challenges related to dataset scarcity, representativeness, and accessibility are discussed alongside recommendations for enhancing dataset creation and utilization. The full dataset inventory is publicly accessible via GitHub and archived on Zenodo, serving as a valuable resource for researchers and developers. By addressing current gaps and proposing future directions, this work aims to support the advancement of Italian speech technologies and linguistic research.
2501.06561
Where to Go Next Day: Multi-scale Spatial-Temporal Decoupled Model for Mid-term Human Mobility Prediction
cs.AI
Predicting individual mobility patterns is crucial across various applications. While current methods mainly focus on predicting the next location for personalized services like recommendations, they often fall short in supporting broader applications such as traffic management and epidemic control, which require longer period forecasts of human mobility. This study addresses mid-term mobility prediction, aiming to capture daily travel patterns and forecast trajectories for the upcoming day or week. We propose a novel Multi-scale Spatial-Temporal Decoupled Predictor (MSTDP) designed to efficiently extract spatial and temporal information by decoupling daily trajectories into distinct location-duration chains. Our approach employs a hierarchical encoder to model multi-scale temporal patterns, including daily recurrence and weekly periodicity, and utilizes a transformer-based decoder to globally attend to predicted information in the location or duration chain. Additionally, we introduce a spatial heterogeneous graph learner to capture multi-scale spatial relationships, enhancing semantic-rich representations. Extensive experiments, including statistical physics analysis, are conducted on large-scale mobile phone records in five cities (Boston, Los Angeles, SF Bay Area, Shanghai, and Tokyo), to demonstrate MSTDP's advantages. Applied to epidemic modeling in Boston, MSTDP significantly outperforms the best-performing baseline, achieving a remarkable 62.8% reduction in MAE for cumulative new cases.
2501.06562
Discrete Speech Unit Extraction via Independent Component Analysis
eess.AS cs.AI cs.LG cs.SD
Self-supervised speech models (S3Ms) have become a common tool for the speech processing community, leveraging representations for downstream tasks. Clustering S3M representations yields discrete speech units (DSUs), which serve as compact representations for speech signals. DSUs are typically obtained by k-means clustering. Using DSUs often leads to strong performance in various tasks, including automatic speech recognition (ASR). However, even with the high dimensionality and redundancy of S3M representations, preprocessing S3M representations for better clustering remains unexplored, even though it can affect the quality of DSUs. In this paper, we investigate the potential of linear preprocessing methods for extracting DSUs. We evaluate standardization, principal component analysis, whitening, and independent component analysis (ICA) on DSU-based ASR benchmarks and demonstrate their effectiveness as preprocessing for k-means. We also conduct extensive analyses of their behavior, such as orthogonality or interpretability of individual components of ICA.
2501.06564
Natural Language Processing and Deep Learning Models to Classify Phase of Flight in Aviation Safety Occurrences
cs.CL cs.LG
The air transport system recognizes the criticality of safety, as even minor anomalies can have severe consequences. Reporting accidents and incidents play a vital role in identifying their causes and proposing safety recommendations. However, the narratives describing pre-accident events are presented in unstructured text that is not easily understood by computer systems. Classifying and categorizing safety occurrences based on these narratives can support informed decision-making by aviation industry stakeholders. In this study, researchers applied natural language processing (NLP) and artificial intelligence (AI) models to process text narratives to classify the flight phases of safety occurrences. The classification performance of two deep learning models, ResNet and sRNN was evaluated, using an initial dataset of 27,000 safety occurrence reports from the NTSB. The results demonstrated good performance, with both models achieving an accuracy exceeding 68%, well above the random guess rate of 14% for a seven-class classification problem. The models also exhibited high precision, recall, and F1 scores. The sRNN model greatly outperformed the simplified ResNet model architecture used in this study. These findings indicate that NLP and deep learning models can infer the flight phase from raw text narratives, enabling effective analysis of safety occurrences.
2501.06566
Cooperative Aerial Robot Inspection Challenge: A Benchmark for Heterogeneous Multi-UAV Planning and Lessons Learned
cs.RO cs.SY eess.SY
We propose the Cooperative Aerial Robot Inspection Challenge (CARIC), a simulation-based benchmark for motion planning algorithms in heterogeneous multi-UAV systems. CARIC features UAV teams with complementary sensors, realistic constraints, and evaluation metrics prioritizing inspection quality and efficiency. It offers a ready-to-use perception-control software stack and diverse scenarios to support the development and evaluation of task allocation and motion planning algorithms. Competitions using CARIC were held at IEEE CDC 2023 and the IROS 2024 Workshop on Multi-Robot Perception and Navigation, attracting innovative solutions from research teams worldwide. This paper examines the top three teams from CDC 2023, analyzing their exploration, inspection, and task allocation strategies while drawing insights into their performance across scenarios. The results highlight the task's complexity and suggest promising directions for future research in cooperative multi-UAV systems.
2501.06570
Aster: Enhancing LSM-structures for Scalable Graph Database
cs.DB
There is a proliferation of applications requiring the management of large-scale, evolving graphs under workloads with intensive graph updates and lookups. Driven by this challenge, we introduce Poly-LSM, a high-performance key-value storage engine for graphs with the following novel techniques: (1) Poly-LSM is embedded with a new design of graph-oriented LSM-tree structure that features a hybrid storage model for concisely and effectively storing graph data. (2) Poly-LSM utilizes an adaptive mechanism to handle edge insertions and deletions on graphs with optimized I/O efficiency. (3) Poly-LSM exploits the skewness of graph data to encode the key-value entries. Building upon this foundation, we further implement Aster, a robust and versatile graph database that supports Gremlin query language facilitating various graph applications. In our experiments, we compared Aster against several mainstream real-world graph databases. The results demonstrate that Aster outperforms all baseline graph databases, especially on large-scale graphs. Notably, on the billion-scale Twitter graph dataset, Aster achieves up to 17x throughput improvement compared to the best-performing baseline graph system.
2501.06571
Active Rule Mining for Multivariate Anomaly Detection in Radio Access Networks
cs.LG cs.AI
Multivariate anomaly detection finds its importance in diverse applications. Despite the existence of many detectors to solve this problem, one cannot simply define why an obtained anomaly inferred by the detector is anomalous. This reasoning is required for network operators to understand the root cause of the anomaly and the remedial action that should be taken to counteract its occurrence. Existing solutions in explainable AI may give cues to features that influence an anomaly, but they do not formulate generalizable rules that can be assessed by a domain expert. Furthermore, not all outliers are anomalous in a business sense. There is an unfulfilled need for a system that can interpret anomalies predicted by a multivariate anomaly detector and map these patterns to actionable rules. This paper aims to fulfill this need by proposing a semi-autonomous anomaly rule miner. The proposed method is applicable to both discrete and time series data and is tailored for radio access network (RAN) anomaly detection use cases. The proposed method is demonstrated in this paper with time series RAN data.
2501.06572
Physics-Informed Neuro-Evolution (PINE): A Survey and Prospects
cs.NE cs.CE cs.LG
Deep learning models trained on finite data lack a complete understanding of the physical world. On the other hand, physics-informed neural networks (PINNs) are infused with such knowledge through the incorporation of mathematically expressible laws of nature into their training loss function. By complying with physical laws, PINNs provide advantages over purely data-driven models in limited-data regimes. This feature has propelled them to the forefront of scientific machine learning, a domain characterized by scarce and costly data. However, the vision of accurate physics-informed learning comes with significant challenges. This review examines PINNs for the first time in terms of model optimization and generalization, shedding light on the need for new algorithmic advances to overcome issues pertaining to the training speed, precision, and generalizability of today's PINN models. Of particular interest are the gradient-free methods of neuroevolution for optimizing the uniquely complex loss landscapes arising in PINN training. Methods synergizing gradient descent and neuroevolution for discovering bespoke neural architectures and balancing multiple conflicting terms in physics-informed learning objectives are positioned as important avenues for future research. Yet another exciting track is to cast neuroevolution as a meta-learner of generalizable PINN models.
2501.06573
Modeling the residual queue and queue-dependent capacity in a static traffic assignment problem
eess.SY cs.SY
The residual queue during a given study period (e.g., peak hour) is an important feature that should be considered when solving a traffic assignment problem under equilibrium for strategic traffic planning. Although studies have focused extensively on static or quasi-dynamic traffic assignment models considering the residual queue, they have failed to capture the situation wherein the equilibrium link flow passing through the link is less than the link physical capacity under congested conditions. To address this critical issue, we introduce a novel static traffic assignment model that explicitly incorporates the residual queue and queue-dependent link capacity. The proposed model ensures that equilibrium link flows remain within the physical capacity bounds, yielding estimations more aligned with data observed by traffic detectors, especially in oversaturated scenarios. A generalized link cost function considering queue-dependent capacity, with an additional queuing delay term is proposed. The queuing delay term represents the added travel cost under congestion, offering a framework wherein conventional static models, both with and without physical capacity constraints, become special cases of our model. Our study rigorously analyzes the mathematical properties of the new model, establishing the theoretical uniqueness of solutions for link flow and residual queue under certain conditions. We also introduce a gradient projection-based alternating minimization algorithm tailored for the proposed model. Numerical examples are conducted to demonstrate the superiority and merit of the proposed model and solution algorithm.
2501.06577
Transforming Social Science Research with Transfer Learning: Social Science Survey Data Integration with AI
cs.AI
Large-N nationally representative surveys, which have profoundly shaped American politics scholarship, represent related but distinct domains -a key condition for transfer learning applications. These surveys are related through their shared demographic, party identification, and ideological variables, yet differ in that individual surveys often lack specific policy preference questions that researchers require. Our study introduces a novel application of transfer learning (TL) to address these gaps, marking the first systematic use of TL paradigms in the context of survey data. Specifically, models pre-trained on the Cooperative Election Study (CES) dataset are fine-tuned for use in the American National Election Studies (ANES) dataset to predict policy questions based on demographic variables. Even with a naive architecture, our transfer learning approach achieves approximately 92 percentage accuracy in predicting missing variables across surveys, demonstrating the robust potential of this method. Beyond this specific application, our paper argues that transfer learning is a promising framework for maximizing the utility of existing survey data. We contend that artificial intelligence, particularly transfer learning, opens new frontiers in social science methodology by enabling systematic knowledge transfer between well-administered surveys that share common variables but differ in their outcomes of interest.
2501.06581
Recommending the right academic programs: An interest mining approach using BERTopic
cs.LG cs.CY cs.IR
Prospective students face the challenging task of selecting a university program that will shape their academic and professional careers. For decision-makers and support services, it is often time-consuming and extremely difficult to match personal interests with suitable programs due to the vast and complex catalogue information available. This paper presents the first information system that provides students with efficient recommendations based on both program content and personal preferences. BERTopic, a powerful topic modeling algorithm, is used that leverages text embedding techniques to generate topic representations. It enables us to mine interest topics from all course descriptions, representing the full body of knowledge taught at the institution. Underpinned by the student's individual choice of topics, a shortlist of the most relevant programs is computed through statistical backtracking in the knowledge map, a novel characterization of the program-course relationship. This approach can be applied to a wide range of educational settings, including professional and vocational training. A case study at a post-secondary school with 80 programs and over 5,000 courses shows that the system provides immediate and effective decision support. The presented interest topics are meaningful, leading to positive effects such as serendipity, personalization, and fairness, as revealed by a qualitative study involving 65 students. Over 98% of users indicated that the recommendations aligned with their interests, and about 94% stated they would use the tool in the future. Quantitative analysis shows the system can be configured to ensure fairness, achieving 98% program coverage while maintaining a personalization score of 0.77. These findings suggest that this real-time, user-centered, data-driven system could improve the program selection process.
2501.06582
ACORD: An Expert-Annotated Retrieval Dataset for Legal Contract Drafting
cs.CL
Information retrieval, specifically contract clause retrieval, is foundational to contract drafting because lawyers rarely draft contracts from scratch; instead, they locate and revise the most relevant precedent. We introduce the Atticus Clause Retrieval Dataset (ACORD), the first retrieval benchmark for contract drafting fully annotated by experts. ACORD focuses on complex contract clauses such as Limitation of Liability, Indemnification, Change of Control, and Most Favored Nation. It includes 114 queries and over 126,000 query-clause pairs, each ranked on a scale from 1 to 5 stars. The task is to find the most relevant precedent clauses to a query. The bi-encoder retriever paired with pointwise LLMs re-rankers shows promising results. However, substantial improvements are still needed to effectively manage the complex legal work typically undertaken by lawyers. As the first retrieval benchmark for contract drafting annotated by experts, ACORD can serve as a valuable IR benchmark for the NLP community.
2501.06583
Optimizing wheel loader performance: an end-to-end approach
cs.CE cs.SY eess.SY
Wheel loaders in mines and construction sites repeatedly load soil from a pile to load receivers. This task presents a challenging optimization problem since each loading's performance depends on the pile state, which depends on previous loadings. We investigate an end-to-end optimization approach considering future loading outcomes and V-cycle transportation costs. To predict the evolution of the pile state and the loading performance, we use world models that leverage deep neural networks trained on numerous simulated loading cycles. A look-ahead tree search optimizes the sequence of loading actions by evaluating the performance of thousands of action candidates, which expand into subsequent action candidates under the predicted pile states recursively. Test results demonstrate that, over a horizon of 15 sequential loadings, the look-ahead tree search is 6% more efficient than a greedy strategy, which always selects the action that maximizes the current single loading performance, and 14% more efficient than using a fixed loading controller optimized for the nominal case.
2501.06585
Boundary-enhanced time series data imputation with long-term dependency diffusion models
cs.LG cs.SI
Data imputation is crucial for addressing challenges posed by missing values in multivariate time series data across various fields, such as healthcare, traffic, and economics, and has garnered significant attention. Among various methods, diffusion model-based approaches show notable performance improvements. However, existing methods often cause disharmonious boundaries between missing and known regions and overlook long-range dependencies in missing data estimation, leading to suboptimal results. To address these issues, we propose a Diffusion-based time Series Data Imputation (DSDI) framework. We develop a weight-reducing injection strategy that incorporates the predicted values of missing points with reducing weights into the reverse diffusion process to mitigate boundary inconsistencies. Further, we introduce a multi-scale S4-based U-Net, which combines hierarchical information from different levels via multi-resolution integration to capture long-term dependencies. Experimental results demonstrate that our model outperforms existing imputation methods.
2501.06588
A Tight VC-Dimension Analysis of Clustering Coresets with Applications
cs.CG cs.DS cs.LG
We consider coresets for $k$-clustering problems, where the goal is to assign points to centers minimizing powers of distances. A popular example is the $k$-median objective $\sum_{p}\min_{c\in C}dist(p,C)$. Given a point set $P$, a coreset $\Omega$ is a small weighted subset that approximates the cost of $P$ for all candidate solutions $C$ up to a $(1\pm\varepsilon )$ multiplicative factor. In this paper, we give a sharp VC-dimension based analysis for coreset construction. As a consequence, we obtain improved $k$-median coreset bounds for the following metrics: Coresets of size $\tilde{O}\left(k\varepsilon^{-2}\right)$ for shortest path metrics in planar graphs, improving over the bounds $\tilde{O}\left(k\varepsilon^{-6}\right)$ by [Cohen-Addad, Saulpic, Schwiegelshohn, STOC'21] and $\tilde{O}\left(k^2\varepsilon^{-4}\right)$ by [Braverman, Jiang, Krauthgamer, Wu, SODA'21]. Coresets of size $\tilde{O}\left(kd\ell\varepsilon^{-2}\log m\right)$ for clustering $d$-dimensional polygonal curves of length at most $m$ with curves of length at most $\ell$ with respect to Frechet metrics, improving over the bounds $\tilde{O}\left(k^3d\ell\varepsilon^{-3}\log m\right)$ by [Braverman, Cohen-Addad, Jiang, Krauthgamer, Schwiegelshohn, Toftrup, and Wu, FOCS'22] and $\tilde{O}\left(k^2d\ell\varepsilon^{-2}\log m \log |P|\right)$ by [Conradi, Kolbe, Psarros, Rohde, SoCG'24].
2501.06589
Ladder-residual: parallelism-aware architecture for accelerating large model inference with communication overlapping
cs.LG cs.CL cs.DC
Large language model inference is both memory-intensive and time-consuming, often requiring distributed algorithms to efficiently scale. Various model parallelism strategies are used in multi-gpu training and inference to partition computation across multiple devices, reducing memory load and computation time. However, using model parallelism necessitates communication of information between GPUs, which has been a major bottleneck and limits the gains obtained by scaling up the number of devices. We introduce Ladder Residual, a simple architectural modification applicable to all residual-based models that enables straightforward overlapping that effectively hides the latency of communication. Our insight is that in addition to systems optimization, one can also redesign the model architecture to decouple communication from computation. While Ladder Residual can allow communication-computation decoupling in conventional parallelism patterns, we focus on Tensor Parallelism in this paper, which is particularly bottlenecked by its heavy communication. For a Transformer model with 70B parameters, applying Ladder Residual to all its layers can achieve 29% end-to-end wall clock speed up at inference time with TP sharding over 8 devices. We refer the resulting Transformer model as the Ladder Transformer. We train a 1B and 3B Ladder Transformer from scratch and observe comparable performance to a standard dense transformer baseline. We also show that it is possible to convert parts of the Llama-3.1 8B model to our Ladder Residual architecture with minimal accuracy degradation by only retraining for 3B tokens. We release our code for training and inference for easier replication of experiments.
2501.06590
ChemAgent: Self-updating Library in Large Language Models Improves Chemical Reasoning
cs.CL cs.AI
Chemical reasoning usually involves complex, multi-step processes that demand precise calculations, where even minor errors can lead to cascading failures. Furthermore, large language models (LLMs) encounter difficulties handling domain-specific formulas, executing reasoning steps accurately, and integrating code effectively when tackling chemical reasoning tasks. To address these challenges, we present ChemAgent, a novel framework designed to improve the performance of LLMs through a dynamic, self-updating library. This library is developed by decomposing chemical tasks into sub-tasks and compiling these sub-tasks into a structured collection that can be referenced for future queries. Then, when presented with a new problem, ChemAgent retrieves and refines pertinent information from the library, which we call memory, facilitating effective task decomposition and the generation of solutions. Our method designs three types of memory and a library-enhanced reasoning component, enabling LLMs to improve over time through experience. Experimental results on four chemical reasoning datasets from SciBench demonstrate that ChemAgent achieves performance gains of up to 46% (GPT-4), significantly outperforming existing methods. Our findings suggest substantial potential for future applications, including tasks such as drug discovery and materials science. Our code can be found at https://github.com/gersteinlab/chemagent
2501.06591
Exploring Pose-Based Anomaly Detection for Retail Security: A Real-World Shoplifting Dataset and Benchmark
cs.CV cs.AI
Shoplifting poses a significant challenge for retailers, resulting in billions of dollars in annual losses. Traditional security measures often fall short, highlighting the need for intelligent solutions capable of detecting shoplifting behaviors in real time. This paper frames shoplifting detection as an anomaly detection problem, focusing on the identification of deviations from typical shopping patterns. We introduce PoseLift, a privacy-preserving dataset specifically designed for shoplifting detection, addressing challenges such as data scarcity, privacy concerns, and model biases. PoseLift is built in collaboration with a retail store and contains anonymized human pose data from real-world scenarios. By preserving essential behavioral information while anonymizing identities, PoseLift balances privacy and utility. We benchmark state-of-the-art pose-based anomaly detection models on this dataset, evaluating performance using a comprehensive set of metrics. Our results demonstrate that pose-based approaches achieve high detection accuracy while effectively addressing privacy and bias concerns inherent in traditional methods. As one of the first datasets capturing real-world shoplifting behaviors, PoseLift offers researchers a valuable tool to advance computer vision ethically and will be publicly available to foster innovation and collaboration. The dataset is available at https://github.com/TeCSAR-UNCC/PoseLift.
2501.06597
EmoXpt: Analyzing Emotional Variances in Human Comments and LLM-Generated Responses
cs.LG cs.CL cs.HC
The widespread adoption of generative AI has generated diverse opinions, with individuals expressing both support and criticism of its applications. This study investigates the emotional dynamics surrounding generative AI by analyzing human tweets referencing terms such as ChatGPT, OpenAI, Copilot, and LLMs. To further understand the emotional intelligence of ChatGPT, we examine its responses to selected tweets, highlighting differences in sentiment between human comments and LLM-generated responses. We introduce EmoXpt, a sentiment analysis framework designed to assess both human perspectives on generative AI and the sentiment embedded in ChatGPT's responses. Unlike prior studies that focus exclusively on human sentiment, EmoXpt uniquely evaluates the emotional expression of ChatGPT. Experimental results demonstrate that LLM-generated responses are notably more efficient, cohesive, and consistently positive than human responses.
2501.06598
ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generation
cs.AI
Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in chart understanding tasks. However, interpreting charts with textual descriptions often leads to information loss, as it fails to fully capture the dense information embedded in charts. In contrast, parsing charts into code provides lossless representations that can effectively contain all critical details. Although existing open-source MLLMs have achieved success in chart understanding tasks, they still face two major challenges when applied to chart-to-code tasks.: (1) Low executability and poor restoration of chart details in the generated code and (2) Lack of large-scale and diverse training data. To address these challenges, we propose \textbf{ChartCoder}, the first dedicated chart-to-code MLLM, which leverages Code LLMs as the language backbone to enhance the executability of the generated code. Furthermore, we introduce \textbf{Chart2Code-160k}, the first large-scale and diverse dataset for chart-to-code generation, and propose the \textbf{Snippet-of-Thought (SoT)} method, which transforms direct chart-to-code generation data into step-by-step generation. Experiments demonstrate that ChartCoder, with only 7B parameters, surpasses existing open-source MLLMs on chart-to-code benchmarks, achieving superior chart restoration and code excitability. Our code will be available at https://github.com/thunlp/ChartCoder.
2501.06602
A Comparative Performance Analysis of Classification and Segmentation Models on Bangladeshi Pothole Dataset
cs.CV
The study involves a comprehensive performance analysis of popular classification and segmentation models, applied over a Bangladeshi pothole dataset, being developed by the authors of this research. This custom dataset of 824 samples, collected from the streets of Dhaka and Bogura performs competitively against the existing industrial and custom datasets utilized in the present literature. The dataset was further augmented four-fold for segmentation and ten-fold for classification evaluation. We tested nine classification models (CCT, CNN, INN, Swin Transformer, ConvMixer, VGG16, ResNet50, DenseNet201, and Xception) and four segmentation models (U-Net, ResU-Net, U-Net++, and Attention-Unet) over both the datasets. Among the classification models, lightweight models namely CCT, CNN, INN, Swin Transformer, and ConvMixer were emphasized due to their low computational requirements and faster prediction times. The lightweight models performed respectfully, oftentimes equating to the performance of heavyweight models. In addition, augmentation was found to enhance the performance of all the tested models. The experimental results exhibit that, our dataset performs on par or outperforms the similar classification models utilized in the existing literature, reaching accuracy and f1-scores over 99%. The dataset also performed on par with the existing datasets for segmentation, achieving model Dice Similarity Coefficient up to 67.54% and IoU scores up to 59.39%.
2501.06603
Preconditioned Sharpness-Aware Minimization: Unifying Analysis and a Novel Learning Algorithm
cs.LG
Targeting solutions over `flat' regions of the loss landscape, sharpness-aware minimization (SAM) has emerged as a powerful tool to improve generalizability of deep neural network based learning. While several SAM variants have been developed to this end, a unifying approach that also guides principled algorithm design has been elusive. This contribution leverages preconditioning (pre) to unify SAM variants and provide not only unifying convergence analysis, but also valuable insights. Building upon preSAM, a novel algorithm termed infoSAM is introduced to address the so-called adversarial model degradation issue in SAM by adjusting gradients depending on noise estimates. Extensive numerical tests demonstrate the superiority of infoSAM across various benchmarks.
2501.06605
RoboHorizon: An LLM-Assisted Multi-View World Model for Long-Horizon Robotic Manipulation
cs.RO
Efficient control in long-horizon robotic manipulation is challenging due to complex representation and policy learning requirements. Model-based visual reinforcement learning (RL) has shown great potential in addressing these challenges but still faces notable limitations, particularly in handling sparse rewards and complex visual features in long-horizon environments. To address these limitations, we propose the Recognize-Sense-Plan-Act (RSPA) pipeline for long-horizon tasks and further introduce RoboHorizon, an LLM-assisted multi-view world model tailored for long-horizon robotic manipulation. In RoboHorizon, pre-trained LLMs generate dense reward structures for multi-stage sub-tasks based on task language instructions, enabling robots to better recognize long-horizon tasks. Keyframe discovery is then integrated into the multi-view masked autoencoder (MAE) architecture to enhance the robot's ability to sense critical task sequences, strengthening its multi-stage perception of long-horizon processes. Leveraging these dense rewards and multi-view representations, a robotic world model is constructed to efficiently plan long-horizon tasks, enabling the robot to reliably act through RL algorithms. Experiments on two representative benchmarks, RLBench and FurnitureBench, show that RoboHorizon outperforms state-of-the-art visual model-based RL methods, achieving a 23.35% improvement in task success rates on RLBench's 4 short-horizon tasks and a 29.23% improvement on 6 long-horizon tasks from RLBench and 3 furniture assembly tasks from FurnitureBench.
2501.06608
Dual-Modality Representation Learning for Molecular Property Prediction
cs.LG q-bio.QM
Molecular property prediction has attracted substantial attention recently. Accurate prediction of drug properties relies heavily on effective molecular representations. The structures of chemical compounds are commonly represented as graphs or SMILES sequences. Recent advances in learning drug properties commonly employ Graph Neural Networks (GNNs) based on the graph representation. For the SMILES representation, Transformer-based architectures have been adopted by treating each SMILES string as a sequence of tokens. Because each representation has its own advantages and disadvantages, combining both representations in learning drug properties is a promising direction. We propose a method named Dual-Modality Cross-Attention (DMCA) that can effectively combine the strengths of two representations by employing the cross-attention mechanism. DMCA was evaluated across eight datasets including both classification and regression tasks. Results show that our method achieves the best overall performance, highlighting its effectiveness in leveraging the complementary information from both graph and SMILES modalities.
2501.06625
Guided Code Generation with LLMs: A Multi-Agent Framework for Complex Code Tasks
cs.AI
Large Language Models (LLMs) have shown remarkable capabilities in code generation tasks, yet they face significant limitations in handling complex, long-context programming challenges and demonstrating complex compositional reasoning abilities. This paper introduces a novel agentic framework for ``guided code generation'' that tries to address these limitations through a deliberately structured, fine-grained approach to code generation tasks. Our framework leverages LLMs' strengths as fuzzy searchers and approximate information retrievers while mitigating their weaknesses in long sequential reasoning and long-context understanding. Empirical evaluation using OpenAI's HumanEval benchmark with Meta's Llama 3.1 8B model (int4 precision) demonstrates a 23.79\% improvement in solution accuracy compared to direct one-shot generation. Our results indicate that structured, guided approaches to code generation can significantly enhance the practical utility of LLMs in software development while overcoming their inherent limitations in compositional reasoning and context handling.
2501.06628
Quantifying Relational Exploration in Cultural Heritage Knowledge Graphs with LLMs: A Neuro-Symbolic Approach
cs.AI
This paper introduces a neuro-symbolic approach for relational exploration in cultural heritage knowledge graphs, leveraging Large Language Models (LLMs) for explanation generation and a novel mathematical framework to quantify the interestingness of relationships. We demonstrate the importance of interestingness measure using a quantitative analysis, by highlighting its impact on the overall performance of our proposed system, particularly in terms of precision, recall, and F1-score. Using the Wikidata Cultural Heritage Linked Open Data (WCH-LOD) dataset, our approach yields a precision of 0.70, recall of 0.68, and an F1-score of 0.69, representing an improvement compared to graph-based (precision: 0.28, recall: 0.25, F1-score: 0.26) and knowledge-based baselines (precision: 0.45, recall: 0.42, F1-score: 0.43). Furthermore, our LLM-powered explanations exhibit better quality, reflected in BLEU (0.52), ROUGE-L (0.58), and METEOR (0.63) scores, all higher than the baseline approaches. We show a strong correlation (0.65) between interestingness measure and the quality of generated explanations, validating its effectiveness. The findings highlight the importance of LLMs and a mathematical formalization for interestingness in enhancing the effectiveness of relational exploration in cultural heritage knowledge graphs, with results that are measurable and testable. We further show that the system enables more effective exploration compared to purely knowledge-based and graph-based methods.
2501.06635
A Reduced Order Iterative Linear Quadratic Regulator (ILQR) Technique for the Optimal Control of Nonlinear Partial Differential Equations
eess.SY cs.SY
In this paper, we introduce a reduced order model-based reinforcement learning (MBRL) approach, utilizing the Iterative Linear Quadratic Regulator (ILQR) algorithm for the optimal control of nonlinear partial differential equations (PDEs). The approach proposes a novel modification of the ILQR technique: it uses the Method of Snapshots to identify a reduced order Linear Time Varying (LTV) approximation of the nonlinear PDE dynamics around a current estimate of the optimal trajectory, utilizes the identified LTV model to solve a time-varying reduced order LQR problem to obtain an improved estimate of the optimal trajectory along with a new reduced basis, and iterates till convergence. The convergence behavior of the reduced order approach is analyzed and the algorithm is shown to converge to a limit set that is dependent on the truncation error in the reduction. The proposed approach is tested on the viscous Burger's equation and two phase-field models for microstructure evolution in materials, and the results show that there is a significant reduction in the computational burden over the standard ILQR approach, without significantly sacrificing performance.
2501.06636
Dual use issues in the field of Natural Language Generation
cs.CL
This report documents the results of a recent survey in the SIGGEN community, focusing on Dual Use issues in Natural Language Generation (NLG). SIGGEN is the Special Interest Group (SIG) of the Association for Computational Linguistics (ACL) for researchers working on NLG. The survey was prompted by the ACL executive board, which asked all SIGs to provide an overview of dual use issues within their respective subfields. The survey was sent out in October 2024 and the results were processed in January 2025. With 23 respondents, the survey is presumably not representative of all SIGGEN members, but at least this document offers a helpful resource for future discussions. This report is open to feedback from the SIGGEN community. Let me know if you have any questions or comments!
2501.06638
Scaling Down Semantic Leakage: Investigating Associative Bias in Smaller Language Models
cs.CL
Semantic leakage is a phenomenon recently introduced by Gonen et al. (2024). It refers to a situation in which associations learnt from the training data emerge in language model generations in an unexpected and sometimes undesired way. Prior work has focused on leakage in large language models (7B+ parameters). In this study, I use Qwen2.5 model family to explore whether smaller models, ranging from 500M to 7B parameters, demonstrate less semantic leakage due to their limited capacity for capturing complex associations. Building on the previous dataset from Gonen et al. (2024), I introduce a new dataset of color-focused prompts, categorized into specific types of semantic associations, to systematically evaluate the models' performance. Results indicate that smaller models exhibit less semantic leakage overall, although this trend is not strictly linear, with medium-sized models sometimes surpassing larger ones in leaking behavior. The dataset, the model generations, and the evaluation code are publicly available at https://github.com/smilni/semantic_leakage_project.
2501.06639
Enhancing Path Planning Performance through Image Representation Learning of High-Dimensional Configuration Spaces
cs.RO cs.AI
This paper presents a novel method for accelerating path-planning tasks in unknown scenes with obstacles by utilizing Wasserstein Generative Adversarial Networks (WGANs) with Gradient Penalty (GP) to approximate the distribution of waypoints for a collision-free path using the Rapidly-exploring Random Tree algorithm. Our approach involves conditioning the WGAN-GP with a forward diffusion process in a continuous latent space to handle multimodal datasets effectively. We also propose encoding the waypoints of a collision-free path as a matrix, where the multidimensional ordering of the waypoints is naturally preserved. This method not only improves model learning but also enhances training convergence. Furthermore, we propose a method to assess whether the trained model fails to accurately capture the true waypoints. In such cases, we revert to uniform sampling to ensure the algorithm's probabilistic completeness; a process that traditionally involves manually determining an optimal ratio for each scenario in other machine learning-based methods. Our experiments demonstrate promising results in accelerating path-planning tasks under critical time constraints. The source code is openly available at https://bitbucket.org/joro3001/imagewgangpplanning/src/master/.
2501.06641
A Permutation-Free Length 3 Decimal Check Digit Code
cs.IT math.CO math.IT
In 1969 J. Verhoeff provided the first examples of a decimal error detecting code using a single check digit to provide protection against all single, transposition and adjacent twin errors. The three codes he presented are length 3-digit codes with 2 information digits. Existence of a 4-digit code would imply the existence of 10 such disjoint 3-digit codes. Apparently, not even a pair of such disjoint 3-digit codes is known. The code developed herein, has the property that the knowledge of any two digits is sufficient to determine the entire codeword even though their positions were unknown. This fulfills Verhoeff's desire to eliminate "cyclic errors". Phonetic errors, where 2 digit pairs of the forms X0 and 1X are interchanged, are also eliminated.
2501.06642
Common Sense Is All You Need
cs.AI
Artificial intelligence (AI) has made significant strides in recent years, yet it continues to struggle with a fundamental aspect of cognition present in all animals: common sense. Current AI systems, including those designed for complex tasks like autonomous driving, problem-solving challenges such as the Abstraction and Reasoning Corpus (ARC), and conversational benchmarks like the Turing Test, often lack the ability to adapt to new situations without extensive prior knowledge. This manuscript argues that integrating common sense into AI systems is essential for achieving true autonomy and unlocking the full societal and commercial value of AI. We propose a shift in the order of knowledge acquisition emphasizing the importance of developing AI systems that start from minimal prior knowledge and are capable of contextual learning, adaptive reasoning, and embodiment -- even within abstract domains. Additionally, we highlight the need to rethink the AI software stack to address this foundational challenge. Without common sense, AI systems may never reach true autonomy, instead exhibiting asymptotic performance that approaches theoretical ideals like AIXI but remains unattainable in practice due to infinite resource and computation requirements. While scaling AI models and passing benchmarks like the Turing Test have brought significant advancements in applications that do not require autonomy, these approaches alone are insufficient to achieve autonomous AI with common sense. By redefining existing benchmarks and challenges to enforce constraints that require genuine common sense, and by broadening our understanding of embodiment to include both physical and abstract domains, we can encourage the development of AI systems better equipped to handle the complexities of real-world and abstract environments.
2501.06645
FocalPO: Enhancing Preference Optimizing by Focusing on Correct Preference Rankings
cs.CL cs.AI
Efficient preference optimization algorithms such as Direct Preference Optimization (DPO) have become a popular approach in aligning large language models (LLMs) with human preferences. These algorithms implicitly treat the LLM as a reward model, and focus on training it to correct misranked preference pairs. However, recent work~\citep{chen2024preference} empirically finds that DPO training \textit{rarely improves these misranked preference pairs}, despite its gradient emphasizing on these cases. We introduce FocalPO, a DPO variant that instead \textit{down-weighs} misranked preference pairs and prioritizes enhancing the model's understanding of pairs that it can already rank correctly. Inspired by Focal Loss used in vision tasks, FocalPO achieves this by adding a modulating factor to dynamically scale DPO loss. Our experiment demonstrates that FocalPO surpasses DPO and its variants on popular benchmarks like Alpaca Eval 2.0 using Mistral-Base-7B and Llama-3-Instruct-8B. Additionally, we empirically reveals how FocalPO affects training on correct and incorrect sample groups, further underscoring its effectiveness.
2501.06650
SafeSplit: A Novel Defense Against Client-Side Backdoor Attacks in Split Learning
cs.CR cs.DC cs.LG
Split Learning (SL) is a distributed deep learning approach enabling multiple clients and a server to collaboratively train and infer on a shared deep neural network (DNN) without requiring clients to share their private local data. The DNN is partitioned in SL, with most layers residing on the server and a few initial layers and inputs on the client side. This configuration allows resource-constrained clients to participate in training and inference. However, the distributed architecture exposes SL to backdoor attacks, where malicious clients can manipulate local datasets to alter the DNN's behavior. Existing defenses from other distributed frameworks like Federated Learning are not applicable, and there is a lack of effective backdoor defenses specifically designed for SL. We present SafeSplit, the first defense against client-side backdoor attacks in Split Learning (SL). SafeSplit enables the server to detect and filter out malicious client behavior by employing circular backward analysis after a client's training is completed, iteratively reverting to a trained checkpoint where the model under examination is found to be benign. It uses a two-fold analysis to identify client-induced changes and detect poisoned models. First, a static analysis in the frequency domain measures the differences in the layer's parameters at the server. Second, a dynamic analysis introduces a novel rotational distance metric that assesses the orientation shifts of the server's layer parameters during training. Our comprehensive evaluation across various data distributions, client counts, and attack scenarios demonstrates the high efficacy of this dual analysis in mitigating backdoor attacks while preserving model utility.
2501.06651
Parking Space Detection in the City of Granada
cs.CV
This paper addresses the challenge of parking space detection in urban areas, focusing on the city of Granada. Utilizing aerial imagery, we develop and apply semantic segmentation techniques to accurately identify parked cars, moving cars and roads. A significant aspect of our research is the creation of a proprietary dataset specific to Granada, which is instrumental in training our neural network model. We employ Fully Convolutional Networks, Pyramid Networks and Dilated Convolutions, demonstrating their effectiveness in urban semantic segmentation. Our approach involves comparative analysis and optimization of various models, including Dynamic U-Net, PSPNet and DeepLabV3+, tailored for the segmentation of aerial images. The study includes a thorough experimentation phase, using datasets such as UDD5 and UAVid, alongside our custom Granada dataset. We evaluate our models using metrics like Foreground Accuracy, Dice Coefficient and Jaccard Index. Our results indicate that DeepLabV3+ offers the most promising performance. We conclude with future directions, emphasizing the need for a dedicated neural network for parked car detection and the potential for application in other urban environments. This work contributes to the fields of urban planning and traffic management, providing insights into efficient utilization of parking spaces through advanced image processing techniques.
2501.06653
Theoretical Characterization of Effect of Masks in Snapshot Compressive Imaging
cs.IT eess.IV math.IT stat.AP
Snapshot compressive imaging (SCI) refers to the recovery of three-dimensional data cubes-such as videos or hyperspectral images-from their two-dimensional projections, which are generated by a special encoding of the data with a mask. SCI systems commonly use binary-valued masks that follow certain physical constraints. Optimizing these masks subject to these constraints is expected to improve system performance. However, prior theoretical work on SCI systems focuses solely on independently and identically distributed (i.i.d.) Gaussian masks, which do not permit such optimization. On the other hand, existing practical mask optimizations rely on computationally intensive joint optimizations that provide limited insight into the role of masks and are expected to be sub-optimal due to the non-convexity and complexity of the optimization. In this paper, we analytically characterize the performance of SCI systems employing binary masks and leverage our analysis to optimize hardware parameters. Our findings provide a comprehensive and fundamental understanding of the role of binary masks - with both independent and dependent elements - and their optimization. We also present simulation results that confirm our theoretical findings and further illuminate different aspects of mask design.
2501.06655
Personalized Preference Fine-tuning of Diffusion Models
cs.LG cs.CV
RLHF techniques like DPO can significantly improve the generation quality of text-to-image diffusion models. However, these methods optimize for a single reward that aligns model generation with population-level preferences, neglecting the nuances of individual users' beliefs or values. This lack of personalization limits the efficacy of these models. To bridge this gap, we introduce PPD, a multi-reward optimization objective that aligns diffusion models with personalized preferences. With PPD, a diffusion model learns the individual preferences of a population of users in a few-shot way, enabling generalization to unseen users. Specifically, our approach (1) leverages a vision-language model (VLM) to extract personal preference embeddings from a small set of pairwise preference examples, and then (2) incorporates the embeddings into diffusion models through cross attention. Conditioning on user embeddings, the text-to-image models are fine-tuned with the DPO objective, simultaneously optimizing for alignment with the preferences of multiple users. Empirical results demonstrate that our method effectively optimizes for multiple reward functions and can interpolate between them during inference. In real-world user scenarios, with as few as four preference examples from a new user, our approach achieves an average win rate of 76\% over Stable Cascade, generating images that more accurately reflect specific user preferences.
2501.06659
TWIX: Automatically Reconstructing Structured Data from Templatized Documents
cs.DB cs.CV
Many documents, that we call templatized documents, are programmatically generated by populating fields in a visual template. Effective data extraction from these documents is crucial to supporting downstream analytical tasks. Current data extraction tools often struggle with complex document layouts, incur high latency and/or cost on large datasets, and often require significant human effort, when extracting tables or values given user-specified fields from documents. The key insight of our tool, TWIX, is to predict the underlying template used to create such documents, modeling the visual and structural commonalities across documents. Data extraction based on this predicted template provides a more principled, accurate, and efficient solution at a low cost. Comprehensive evaluations on 34 diverse real-world datasets show that uncovering the template is crucial for data extraction from templatized documents. TWIX achieves over 90% precision and recall on average, outperforming tools from industry: Textract and Azure Document Intelligence, and vision-based LLMs like GPT-4-Vision, by over 25% in precision and recall. TWIX scales easily to large datasets and is 734X faster and 5836X cheaper than vision-based LLMs for extracting data from a large document collection with 817 pages.
2501.06660
MapGS: Generalizable Pretraining and Data Augmentation for Online Mapping via Novel View Synthesis
cs.CV cs.RO
Online mapping reduces the reliance of autonomous vehicles on high-definition (HD) maps, significantly enhancing scalability. However, recent advancements often overlook cross-sensor configuration generalization, leading to performance degradation when models are deployed on vehicles with different camera intrinsics and extrinsics. With the rapid evolution of novel view synthesis methods, we investigate the extent to which these techniques can be leveraged to address the sensor configuration generalization challenge. We propose a novel framework leveraging Gaussian splatting to reconstruct scenes and render camera images in target sensor configurations. The target config sensor data, along with labels mapped to the target config, are used to train online mapping models. Our proposed framework on the nuScenes and Argoverse 2 datasets demonstrates a performance improvement of 18% through effective dataset augmentation, achieves faster convergence and efficient training, and exceeds state-of-the-art performance when using only 25% of the original training data. This enables data reuse and reduces the need for laborious data labeling. Project page at https://henryzhangzhy.github.io/mapgs.
2501.06661
Learning dynamical systems with hit-and-run random feature maps
cs.LG physics.data-an stat.ME stat.ML
We show how random feature maps can be used to forecast dynamical systems with excellent forecasting skill. We consider the tanh activation function and judiciously choose the internal weights in a data-driven manner such that the resulting features explore the nonlinear, non-saturated regions of the activation function. We introduce skip connections and construct a deep variant of random feature maps by combining several units. To mitigate the curse of dimensionality, we introduce localization where we learn local maps, employing conditional independence. Our modified random feature maps provide excellent forecasting skill for both single trajectory forecasts as well as long-time estimates of statistical properties, for a range of chaotic dynamical systems with dimensions up to 512. In contrast to other methods such as reservoir computers which require extensive hyperparameter tuning, we effectively need to tune only a single hyperparameter, and are able to achieve state-of-the-art forecast skill with much smaller networks.
2501.06662
The Magnitude of Categories of Texts Enriched by Language Models
math.CT cs.CL
The purpose of this article is twofold. Firstly, we use the next-token probabilities given by a language model to explicitly define a $[0,1]$-enrichment of a category of texts in natural language, in the sense of Bradley, Terilla, and Vlassopoulos. We consider explicitly the terminating conditions for text generation and determine when the enrichment itself can be interpreted as a probability over texts. Secondly, we compute the M\"obius function and the magnitude of an associated generalized metric space $\mathcal{M}$ of texts using a combinatorial version of these quantities recently introduced by Vigneaux. The magnitude function $f(t)$ of $\mathcal{M}$ is a sum over texts $x$ (prompts) of the Tsallis $t$-entropies of the next-token probability distributions $p(-|x)$ plus the cardinality of the model's possible outputs. The derivative of $f$ at $t=1$ recovers a sum of Shannon entropies, which justifies seeing magnitude as a partition function. Following Leinster and Schulman, we also express the magnitude function of $\mathcal M$ as an Euler characteristic of magnitude homology and provide an explicit description of the zeroeth and first magnitude homology groups.
2501.06663
Ultra Memory-Efficient On-FPGA Training of Transformers via Tensor-Compressed Optimization
cs.LG cs.AR cs.CL
Transformer models have achieved state-of-the-art performance across a wide range of machine learning tasks. There is growing interest in training transformers on resource-constrained edge devices due to considerations such as privacy, domain adaptation, and on-device scientific machine learning. However, the significant computational and memory demands required for transformer training often exceed the capabilities of an edge device. Leveraging low-rank tensor compression, this paper presents the first on-FPGA accelerator for end-to-end transformer training. On the algorithm side, we present a bi-directional contraction flow for tensorized transformer training, significantly reducing the computational FLOPS and intra-layer memory costs compared to existing tensor operations. On the hardware side, we store all highly compressed model parameters and gradient information on chip, creating an on-chip-memory-only framework for each stage in training. This reduces off-chip communication and minimizes latency and energy costs. Additionally, we implement custom computing kernels for each training stage and employ intra-layer parallelism and pipe-lining to further enhance run-time and memory efficiency. Through experiments on transformer models within $36.7$ to $93.5$ MB using FP-32 data formats on the ATIS dataset, our tensorized FPGA accelerator could conduct single-batch end-to-end training on the AMD Alevo U50 FPGA, with a memory budget of less than $6$-MB BRAM and $22.5$-MB URAM. Compared to uncompressed training on the NVIDIA RTX 3090 GPU, our on-FPGA training achieves a memory reduction of $30\times$ to $51\times$. Our FPGA accelerator also achieves up to $3.6\times$ less energy cost per epoch compared with tensor Transformer training on an NVIDIA RTX 3090 GPU.
2501.06669
Challenging reaction prediction models to generalize to novel chemistry
cs.LG physics.chem-ph
Deep learning models for anticipating the products of organic reactions have found many use cases, including validating retrosynthetic pathways and constraining synthesis-based molecular design tools. Despite compelling performance on popular benchmark tasks, strange and erroneous predictions sometimes ensue when using these models in practice. The core issue is that common benchmarks test models in an in-distribution setting, whereas many real-world uses for these models are in out-of-distribution settings and require a greater degree of extrapolation. To better understand how current reaction predictors work in out-of-distribution domains, we report a series of more challenging evaluations of a prototypical SMILES-based deep learning model. First, we illustrate how performance on randomly sampled datasets is overly optimistic compared to performance when generalizing to new patents or new authors. Second, we conduct time splits that evaluate how models perform when tested on reactions published in years after those in their training set, mimicking real-world deployment. Finally, we consider extrapolation across reaction classes to reflect what would be required for the discovery of novel reaction types. This panel of tasks can reveal the capabilities and limitations of today's reaction predictors, acting as a crucial first step in the development of tomorrow's next-generation models capable of reaction discovery.
2501.06670
A Geometric Analysis-Based Safety Assessment Framework for MASS Route Decision-Making in Restricted Waters
eess.SY cs.SY
To enhance the safety of Maritime Autonomous Surface Ships (MASS) navigating in restricted waters, this paper aims to develop a geometric analysis-based route safety assessment (GARSA) framework, specifically designed for their route decision-making in irregularly shaped waterways. Utilizing line and point geometric elements to define waterway boundaries, the framework enables to construct a dynamic width characterization function to quantify spatial safety along intricate waterways. An iterative method is developed to calculate this function, enabling an abstracted spatial property representation of the waterways. Based on this, we introduce a navigational safety index that balances global navigational safety and local risk to determine the safest route. To accommodate ship kinematic constraints, path modifications are applied using a dynamic window approach. A case study in a simulated Port of Hamburg environment shows that GARSA effectively identifies safe routes and avoids the risk of entering narrow waterways in an autonomous manner, thereby prioritizing safety in route decision-making for MASS in confined waters.
2501.06678
Imbalanced Medical Image Segmentation with Pixel-dependent Noisy Labels
cs.CV cs.AI
Accurate medical image segmentation is often hindered by noisy labels in training data, due to the challenges of annotating medical images. Prior research works addressing noisy labels tend to make class-dependent assumptions, overlooking the pixel-dependent nature of most noisy labels. Furthermore, existing methods typically apply fixed thresholds to filter out noisy labels, risking the removal of minority classes and consequently degrading segmentation performance. To bridge these gaps, our proposed framework, Collaborative Learning with Curriculum Selection (CLCS), addresses pixel-dependent noisy labels with class imbalance. CLCS advances the existing works by i) treating noisy labels as pixel-dependent and addressing them through a collaborative learning framework, and ii) employing a curriculum dynamic thresholding approach adapting to model learning progress to select clean data samples to mitigate the class imbalance issue, and iii) applying a noise balance loss to noisy data samples to improve data utilization instead of discarding them outright. Specifically, our CLCS contains two modules: Curriculum Noisy Label Sample Selection (CNS) and Noise Balance Loss (NBL). In the CNS module, we designed a two-branch network with discrepancy loss for collaborative learning so that different feature representations of the same instance could be extracted from distinct views and used to vote the class probabilities of pixels. Besides, a curriculum dynamic threshold is adopted to select clean-label samples through probability voting. In the NBL module, instead of directly dropping the suspiciously noisy labels, we further adopt a robust loss to leverage such instances to boost the performance.
2501.06679
Coordinated Deliverable Energy Flexibility from EV Aggregators in Distribution Networks
eess.SY cs.SY
This paper presents a coordinated framework to optimize electric vehicle (EV) charging considering grid constraints and system uncertainties. The proposed framework consists of two optimization models. In particular, the distribution system operator (DSO) solves the first model to optimize the amount of deliverable energy flexibility that can be obtained from EV aggregators. To address the uncertainties of loads and solar energy generation, a hybrid robust/stochastic approach is employed, enabling the transformation of uncertainty-related constraints into a set of equivalent deterministic constraints. Once the DSO has computed the optimal energy flexibility, each aggregator utilizes the second optimization model to optimize the charging schedule for its respective fleet of EVs. Numerical simulations are performed on a modified IEEE 33-bus distribution network to illustrate the efficiency of the proposed framework.
2501.06680
Application of Vision-Language Model to Pedestrians Behavior and Scene Understanding in Autonomous Driving
cs.CV cs.AI cs.LG cs.RO
Autonomous driving (AD) has experienced significant improvements in recent years and achieved promising 3D detection, classification, and localization results. However, many challenges remain, e.g. semantic understanding of pedestrians' behaviors, and downstream handling for pedestrian interactions. Recent studies in applications of Large Language Models (LLM) and Vision-Language Models (VLM) have achieved promising results in scene understanding and high-level maneuver planning in diverse traffic scenarios. However, deploying the billion-parameter LLMs to vehicles requires significant computation and memory resources. In this paper, we analyzed effective knowledge distillation of semantic labels to smaller Vision networks, which can be used for the semantic representation of complex scenes for downstream decision-making for planning and control.
2501.06682
Generative AI in Education: From Foundational Insights to the Socratic Playground for Learning
cs.AI
This paper explores the synergy between human cognition and Large Language Models (LLMs), highlighting how generative AI can drive personalized learning at scale. We discuss parallels between LLMs and human cognition, emphasizing both the promise and new perspectives on integrating AI systems into education. After examining challenges in aligning technology with pedagogy, we review AutoTutor-one of the earliest Intelligent Tutoring Systems (ITS)-and detail its successes, limitations, and unfulfilled aspirations. We then introduce the Socratic Playground, a next-generation ITS that uses advanced transformer-based models to overcome AutoTutor's constraints and provide personalized, adaptive tutoring. To illustrate its evolving capabilities, we present a JSON-based tutoring prompt that systematically guides learner reflection while tracking misconceptions. Throughout, we underscore the importance of placing pedagogy at the forefront, ensuring that technology's power is harnessed to enhance teaching and learning rather than overshadow it.
2501.06685
Tab-Shapley: Identifying Top-k Tabular Data Quality Insights
cs.LG stat.ML
We present an unsupervised method for aggregating anomalies in tabular datasets by identifying the top-k tabular data quality insights. Each insight consists of a set of anomalous attributes and the corresponding subsets of records that serve as evidence to the user. The process of identifying these insight blocks is challenging due to (i) the absence of labeled anomalies, (ii) the exponential size of the subset search space, and (iii) the complex dependencies among attributes, which obscure the true sources of anomalies. Simple frequency-based methods fail to capture these dependencies, leading to inaccurate results. To address this, we introduce Tab-Shapley, a cooperative game theory based framework that uses Shapley values to quantify the contribution of each attribute to the data's anomalous nature. While calculating Shapley values typically requires exponential time, we show that our game admits a closed-form solution, making the computation efficient. We validate the effectiveness of our approach through empirical analysis on real-world tabular datasets with ground-truth anomaly labels.
2501.06686
Understanding and Mitigating Membership Inference Risks of Neural Ordinary Differential Equations
cs.CR cs.LG
Neural ordinary differential equations (NODEs) are an emerging paradigm in scientific computing for modeling dynamical systems. By accurately learning underlying dynamics in data in the form of differential equations, NODEs have been widely adopted in various domains, such as healthcare, finance, computer vision, and language modeling. However, there remains a limited understanding of the privacy implications of these fundamentally different models, particularly with regard to their membership inference risks. In this work, we study the membership inference risks associated with NODEs. We first comprehensively evaluate NODEs against membership inference attacks. We show that NODEs are twice as resistant to these privacy attacks compared to conventional feedforward models such as ResNets. By analyzing the variance in membership risks across different NODE models, we identify the factors that contribute to their lower risks. We then demonstrate, both theoretically and empirically, that membership inference risks can be further mitigated by utilizing a stochastic variant of NODEs: Neural stochastic differential equations (NSDEs). We show that NSDEs are differentially-private (DP) learners that provide the same provable privacy guarantees as DP-SGD, the de-facto mechanism for training private models. NSDEs are also effective in mitigating existing membership inference attacks, demonstrating risks comparable to private models trained with DP-SGD while offering an improved privacy-utility trade-off. Moreover, we propose a drop-in-replacement strategy that efficiently integrates NSDEs into conventional feedforward models to enhance their privacy.
2501.06689
TAPO: Task-Referenced Adaptation for Prompt Optimization
cs.CL
Prompt engineering can significantly improve the performance of large language models (LLMs), with automated prompt optimization (APO) gaining significant attention due to the time-consuming and laborious nature of manual prompt design. However, much of the existing work in APO overlooks task-specific characteristics, resulting in prompts that lack domain specificity and are not well-suited for task-specific optimization. In this paper, we introduce TAPO, a multitask-aware prompt optimization framework composed of three key modules. First, a task-aware metric selection module is proposed to enhance task-specific prompt generation capabilities. Second, we present a multi-metrics evaluation module to jointly evaluate prompts from multiple perspectives. Third, an evolution-based optimization framework is introduced for automatic prompt refinement, which improves adaptability across various tasks. Extensive experiments on six datasets demonstrate the effectiveness of our approach, and our code is publicly available.
2501.06692
PGP-SAM: Prototype-Guided Prompt Learning for Efficient Few-Shot Medical Image Segmentation
cs.CV cs.AI
The Segment Anything Model (SAM) has demonstrated strong and versatile segmentation capabilities, along with intuitive prompt-based interactions. However, customizing SAM for medical image segmentation requires massive amounts of pixel-level annotations and precise point- or box-based prompt designs. To address these challenges, we introduce PGP-SAM, a novel prototype-based few-shot tuning approach that uses limited samples to replace tedious manual prompts. Our key idea is to leverage inter- and intra-class prototypes to capture class-specific knowledge and relationships. We propose two main components: (1) a plug-and-play contextual modulation module that integrates multi-scale information, and (2) a class-guided cross-attention mechanism that fuses prototypes and features for automatic prompt generation. Experiments on a public multi-organ dataset and a private ventricle dataset demonstrate that PGP-SAM achieves superior mean Dice scores compared with existing prompt-free SAM variants, while using only 10\% of the 2D slices.
2501.06693
Vid2Sim: Realistic and Interactive Simulation from Video for Urban Navigation
cs.CV cs.RO
Sim-to-real gap has long posed a significant challenge for robot learning in simulation, preventing the deployment of learned models in the real world. Previous work has primarily focused on domain randomization and system identification to mitigate this gap. However, these methods are often limited by the inherent constraints of the simulation and graphics engines. In this work, we propose Vid2Sim, a novel framework that effectively bridges the sim2real gap through a scalable and cost-efficient real2sim pipeline for neural 3D scene reconstruction and simulation. Given a monocular video as input, Vid2Sim can generate photorealistic and physically interactable 3D simulation environments to enable the reinforcement learning of visual navigation agents in complex urban environments. Extensive experiments demonstrate that Vid2Sim significantly improves the performance of urban navigation in the digital twins and real world by 31.2% and 68.3% in success rate compared with agents trained with prior simulation methods.
2501.06695
DVM: Towards Controllable LLM Agents in Social Deduction Games
cs.AI
Large Language Models (LLMs) have advanced the capability of game agents in social deduction games (SDGs). These games rely heavily on conversation-driven interactions and require agents to infer, make decisions, and express based on such information. While this progress leads to more sophisticated and strategic non-player characters (NPCs) in SDGs, there exists a need to control the proficiency of these agents. This control not only ensures that NPCs can adapt to varying difficulty levels during gameplay, but also provides insights into the safety and fairness of LLM agents. In this paper, we present DVM, a novel framework for developing controllable LLM agents for SDGs, and demonstrate its implementation on one of the most popular SDGs, Werewolf. DVM comprises three main components: Predictor, Decider, and Discussor. By integrating reinforcement learning with a win rate-constrained decision chain reward mechanism, we enable agents to dynamically adjust their gameplay proficiency to achieve specified win rates. Experiments show that DVM not only outperforms existing methods in the Werewolf game, but also successfully modulates its performance levels to meet predefined win rate targets. These results pave the way for LLM agents' adaptive and balanced gameplay in SDGs, opening new avenues for research in controllable game agents.
2501.06697
Mamba-MOC: A Multicategory Remote Object Counting via State Space Model
cs.CV cs.AI
Multicategory remote object counting is a fundamental task in computer vision, aimed at accurately estimating the number of objects of various categories in remote images. Existing methods rely on CNNs and Transformers, but CNNs struggle to capture global dependencies, and Transformers are computationally expensive, which limits their effectiveness in remote applications. Recently, Mamba has emerged as a promising solution in the field of computer vision, offering a linear complexity for modeling global dependencies. To this end, we propose Mamba-MOC, a mamba-based network designed for multi-category remote object counting, which represents the first application of Mamba to remote sensing object counting. Specifically, we propose a cross-scale interaction module to facilitate the deep integration of hierarchical features. Then we design a context state space model to capture both global and local contextual information and provide local neighborhood information during the scan process. Experimental results in large-scale realistic scenarios demonstrate that our proposed method achieves state-of-the-art performance compared with some mainstream counting algorithms.
2501.06699
Large Language Models, Knowledge Graphs and Search Engines: A Crossroads for Answering Users' Questions
cs.AI cs.IR cs.SC
Much has been discussed about how Large Language Models, Knowledge Graphs and Search Engines can be combined in a synergistic manner. A dimension largely absent from current academic discourse is the user perspective. In particular, there remain many open questions regarding how best to address the diverse information needs of users, incorporating varying facets and levels of difficulty. This paper introduces a taxonomy of user information needs, which guides us to study the pros, cons and possible synergies of Large Language Models, Knowledge Graphs and Search Engines. From this study, we derive a roadmap for future research.
2501.06700
Average Reward Reinforcement Learning for Wireless Radio Resource Management
cs.IT cs.LG cs.NI eess.SP math.IT
In this paper, we address a crucial but often overlooked issue in applying reinforcement learning (RL) to radio resource management (RRM) in wireless communications: the mismatch between the discounted reward RL formulation and the undiscounted goal of wireless network optimization. To the best of our knowledge, we are the first to systematically investigate this discrepancy, starting with a discussion of the problem formulation followed by simulations that quantify the extent of the gap. To bridge this gap, we introduce the use of average reward RL, a method that aligns more closely with the long-term objectives of RRM. We propose a new method called the Average Reward Off policy Soft Actor Critic (ARO SAC) is an adaptation of the well known Soft Actor Critic algorithm in the average reward framework. This new method achieves significant performance improvement our simulation results demonstrate a 15% gain in the system performance over the traditional discounted reward RL approach, underscoring the potential of average reward RL in enhancing the efficiency and effectiveness of wireless network optimization.
2501.06701
Sequential Portfolio Selection under Latent Side Information-Dependence Structure: Optimality and Universal Learning Algorithms
q-fin.MF cs.IT cs.LG math.IT math.PR q-fin.PM
This paper investigates the investment problem of constructing an optimal no-short sequential portfolio strategy in a market with a latent dependence structure between asset prices and partly unobservable side information, which is often high-dimensional. The results demonstrate that a dynamic strategy, which forms a portfolio based on perfect knowledge of the dependence structure and full market information over time, may not grow at a higher rate infinitely often than a constant strategy, which remains invariant over time. Specifically, if the market is stationary, implying that the dependence structure is statistically stable, the growth rate of an optimal dynamic strategy, utilizing the maximum capacity of the entire market information, almost surely decays over time into an equilibrium state, asymptotically converging to the growth rate of a constant strategy. Technically, this work reassesses the common belief that a constant strategy only attains the optimal limiting growth rate of dynamic strategies when the market process is identically and independently distributed. By analyzing the dynamic log-optimal portfolio strategy as the optimal benchmark in a stationary market with side information, we show that a random optimal constant strategy almost surely exists, even when a limiting growth rate for the dynamic strategy does not. Consequently, two approaches to learning algorithms for portfolio construction are discussed, demonstrating the safety of removing side information from the learning process while still guaranteeing an asymptotic growth rate comparable to that of the optimal dynamic strategy.
2501.06704
Fine-tuning ChatGPT for Automatic Scoring of Written Scientific Explanations in Chinese
cs.AI cs.CL
The development of explanations for scientific phenomena is essential in science assessment, but scoring student-written explanations remains challenging and resource-intensive. Large language models (LLMs) have shown promise in addressing this issue, particularly in alphabetic languages like English. However, their applicability to logographic languages is less explored. This study investigates the potential of fine-tuning ChatGPT, a leading LLM, to automatically score scientific explanations written in Chinese. Student responses to seven scientific explanation tasks were collected and automatically scored, with scoring accuracy examined in relation to reasoning complexity using the Kendall correlation. A qualitative analysis explored how linguistic features influenced scoring accuracy. The results show that domain-specific adaptation enables ChatGPT to score Chinese scientific explanations with accuracy. However, scoring accuracy correlates with reasoning complexity: a negative correlation for lower-level responses and a positive one for higher-level responses. The model overrates complex reasoning in low-level responses with intricate sentence structures and underrates high-level responses using concise causal reasoning. These correlations stem from linguistic features--simplicity and clarity enhance accuracy for lower-level responses, while comprehensiveness improves accuracy for higher-level ones. Simpler, shorter responses tend to score more accurately at lower levels, whereas longer, information-rich responses yield better accuracy at higher levels. These findings demonstrate the effectiveness of LLMs in automatic scoring within a Chinese context and emphasize the importance of linguistic features and reasoning complexity in fine-tuning scoring models for educational assessments.
2501.06705
Quantum Data Sketches
cs.DB quant-ph
Recent advancements in quantum technologies, particularly in quantum sensing and simulation, have facilitated the generation and analysis of inherently quantum data. This progress underscores the necessity for developing efficient and scalable quantum data management strategies. This goal faces immense challenges due to the exponential dimensionality of quantum data and its unique quantum properties such as no-cloning and measurement stochasticity. Specifically, classical storage and manipulation of an arbitrary n-qubit quantum state requires exponential space and time. Hence, there is a critical need to revisit foundational data management concepts and algorithms for quantum data. In this paper, we propose succinct quantum data sketches to support basic database operations such as search and selection. We view our work as an initial step towards the development of quantum data management model, opening up many possibilities for future research in this direction.
2501.06706
AIOpsLab: A Holistic Framework to Evaluate AI Agents for Enabling Autonomous Clouds
cs.AI cs.DC cs.MA cs.SE
AI for IT Operations (AIOps) aims to automate complex operational tasks, such as fault localization and root cause analysis, to reduce human workload and minimize customer impact. While traditional DevOps tools and AIOps algorithms often focus on addressing isolated operational tasks, recent advances in Large Language Models (LLMs) and AI agents are revolutionizing AIOps by enabling end-to-end and multitask automation. This paper envisions a future where AI agents autonomously manage operational tasks throughout the entire incident lifecycle, leading to self-healing cloud systems, a paradigm we term AgentOps. Realizing this vision requires a comprehensive framework to guide the design, development, and evaluation of these agents. To this end, we present AIOPSLAB, a framework that not only deploys microservice cloud environments, injects faults, generates workloads, and exports telemetry data but also orchestrates these components and provides interfaces for interacting with and evaluating agents. We discuss the key requirements for such a holistic framework and demonstrate how AIOPSLAB can facilitate the evaluation of next-generation AIOps agents. Through evaluations of state-of-the-art LLM agents within the benchmark created by AIOPSLAB, we provide insights into their capabilities and limitations in handling complex operational tasks in cloud environments.
2501.06707
ELIZA Reanimated: The world's first chatbot restored on the world's first time sharing system
cs.AI cs.CY cs.SC
ELIZA, created by Joseph Weizenbaum at MIT in the early 1960s, is usually considered the world's first chatbot. It was developed in MAD-SLIP on MIT's CTSS, the world's first time-sharing system, on an IBM 7094. We discovered an original ELIZA printout in Prof. Weizenbaum's archives at MIT, including an early version of the famous DOCTOR script, a nearly complete version of the MAD-SLIP code, and various support functions in MAD and FAP. Here we describe the reanimation of this original ELIZA on a restored CTSS, itself running on an emulated IBM 7094. The entire stack is open source, so that any user of a unix-like OS can run the world's first chatbot on the world's first time-sharing system.
2501.06708
Evaluating Sample Utility for Data Selection by Mimicking Model Weights
cs.LG cs.AI
Foundation models are trained on large-scale web-crawled datasets, which often contain noise, biases, and irrelevant information. This motivates the use of data selection techniques, which can be divided into model-free variants -- relying on heuristic rules and downstream datasets -- and model-based, e.g., using influence functions. The former can be expensive to design and risk introducing unwanted dependencies, while the latter are often computationally prohibitive. Instead, we propose an efficient, model-based approach using the Mimic Score, a new data quality metric that leverages the weights of a reference model to assess the usefulness of individual samples for training a new model. It relies on the alignment between gradients and a target direction induced by the reference model. Using the derived Mimic Scores, we develop Grad-Mimic, a framework that prioritizes samples for learning, creates effective filters, and automates data selection. Empirically, using Mimic Scores to guide training improves data efficiency, results in consistent performance gains across six image datasets, and includes enhancements to CLIP models. Moreover, Mimic Score-based filters improve upon existing filtering methods, e.g., cutting 4.7 million samples to train better CLIP models while offering accurate estimation of training dataset quality.
2501.06710
Multi-task Visual Grounding with Coarse-to-Fine Consistency Constraints
cs.CV cs.AI
Multi-task visual grounding involves the simultaneous execution of localization and segmentation in images based on textual expressions. The majority of advanced methods predominantly focus on transformer-based multimodal fusion, aiming to extract robust multimodal representations. However, ambiguity between referring expression comprehension (REC) and referring image segmentation (RIS) is error-prone, leading to inconsistencies between multi-task predictions. Besides, insufficient multimodal understanding directly contributes to biased target perception. To overcome these challenges, we propose a Coarse-to-fine Consistency Constraints Visual Grounding architecture ($\text{C}^3\text{VG}$), which integrates implicit and explicit modeling approaches within a two-stage framework. Initially, query and pixel decoders are employed to generate preliminary detection and segmentation outputs, a process referred to as the Rough Semantic Perception (RSP) stage. These coarse predictions are subsequently refined through the proposed Mask-guided Interaction Module (MIM) and a novel explicit bidirectional consistency constraint loss to ensure consistent representations across tasks, which we term the Refined Consistency Interaction (RCI) stage. Furthermore, to address the challenge of insufficient multimodal understanding, we leverage pre-trained models based on visual-linguistic fusion representations. Empirical evaluations on the RefCOCO, RefCOCO+, and RefCOCOg datasets demonstrate the efficacy and soundness of $\text{C}^3\text{VG}$, which significantly outperforms state-of-the-art REC and RIS methods by a substantial margin. Code and model will be available at \url{https://github.com/Dmmm1997/C3VG}.
2501.06713
MiniRAG: Towards Extremely Simple Retrieval-Augmented Generation
cs.AI
The growing demand for efficient and lightweight Retrieval-Augmented Generation (RAG) systems has highlighted significant challenges when deploying Small Language Models (SLMs) in existing RAG frameworks. Current approaches face severe performance degradation due to SLMs' limited semantic understanding and text processing capabilities, creating barriers for widespread adoption in resource-constrained scenarios. To address these fundamental limitations, we present MiniRAG, a novel RAG system designed for extreme simplicity and efficiency. MiniRAG introduces two key technical innovations: (1) a semantic-aware heterogeneous graph indexing mechanism that combines text chunks and named entities in a unified structure, reducing reliance on complex semantic understanding, and (2) a lightweight topology-enhanced retrieval approach that leverages graph structures for efficient knowledge discovery without requiring advanced language capabilities. Our extensive experiments demonstrate that MiniRAG achieves comparable performance to LLM-based methods even when using SLMs while requiring only 25\% of the storage space. Additionally, we contribute a comprehensive benchmark dataset for evaluating lightweight RAG systems under realistic on-device scenarios with complex queries. We fully open-source our implementation and datasets at: https://github.com/HKUDS/MiniRAG.
2501.06714
F3D-Gaus: Feed-forward 3D-aware Generation on ImageNet with Cycle-Consistent Gaussian Splatting
cs.CV
This paper tackles the problem of generalizable 3D-aware generation from monocular datasets, e.g., ImageNet. The key challenge of this task is learning a robust 3D-aware representation without multi-view or dynamic data, while ensuring consistent texture and geometry across different viewpoints. Although some baseline methods are capable of 3D-aware generation, the quality of the generated images still lags behind state-of-the-art 2D generation approaches, which excel in producing high-quality, detailed images. To address this severe limitation, we propose a novel feed-forward pipeline based on pixel-aligned Gaussian Splatting, coined as F3D-Gaus, which can produce more realistic and reliable 3D renderings from monocular inputs. In addition, we introduce a self-supervised cycle-consistent constraint to enforce cross-view consistency in the learned 3D representation. This training strategy naturally allows aggregation of multiple aligned Gaussian primitives and significantly alleviates the interpolation limitations inherent in single-view pixel-aligned Gaussian Splatting. Furthermore, we incorporate video model priors to perform geometry-aware refinement, enhancing the generation of fine details in wide-viewpoint scenarios and improving the model's capability to capture intricate 3D textures. Extensive experiments demonstrate that our approach not only achieves high-quality, multi-view consistent 3D-aware generation from monocular datasets, but also significantly improves training and inference efficiency.
2501.06715
ZNO-Eval: Benchmarking reasoning capabilities of large language models in Ukrainian
cs.CL cs.AI
As the usage of large language models for problems outside of simple text understanding or generation increases, assessing their abilities and limitations becomes crucial. While significant progress has been made in this area over the last few years, most research has focused on benchmarking English, leaving other languages underexplored. This makes evaluating the reasoning and robustness level of language models in Ukrainian particularly challenging. The purpose of this work is to establish a comprehensive benchmark for the reasoning capabilities evaluation of large language models in the Ukrainian language. This paper presents the ZNO-Eval benchmark based on real exam tasks from Ukraine's standardized educational testing system: the External Independent Evaluation and the National Multi-subject Test. With single-answer options, multiple-choice, matching, and open-ended questions from diverse subjects, including Ukrainian language, mathematics, history, and geography, this dataset paves the way toward a thorough analysis of reasoning capabilities across different domains and complexities. Evaluation of several well-known language models, such as GPT-3.5-Turbo, GPT-4o, GPT-4-Turbo, Mistral Large, Claude 3 Opus, and Gemini-1.5 Pro on this benchmark demonstrated the superiority of GPT-4o in both common knowledge reasoning and intricate language tasks. At the same time, Gemini Pro and GPT-4 Turbo excelled in the arithmetic domain, leading in single-answer and open-ended math problems. While all models were close to max performance in text-only common knowledge tasks like history and geography, there still is a gap for Ukrainian language and math, thus highlighting the importance of developing specialized language benchmarks for more accurate assessments of model capabilities and limitations across different languages and contexts.
2501.06718
DRDT3: Diffusion-Refined Decision Test-Time Training Model
cs.LG
Decision Transformer (DT), a trajectory modeling method, has shown competitive performance compared to traditional offline reinforcement learning (RL) approaches on various classic control tasks. However, it struggles to learn optimal policies from suboptimal, reward-labeled trajectories. In this study, we explore the use of conditional generative modeling to facilitate trajectory stitching given its high-quality data generation ability. Additionally, recent advancements in Recurrent Neural Networks (RNNs) have shown their linear complexity and competitive sequence modeling performance over Transformers. We leverage the Test-Time Training (TTT) layer, an RNN that updates hidden states during testing, to model trajectories in the form of DT. We introduce a unified framework, called Diffusion-Refined Decision TTT (DRDT3), to achieve performance beyond DT models. Specifically, we propose the Decision TTT (DT3) module, which harnesses the sequence modeling strengths of both self-attention and the TTT layer to capture recent contextual information and make coarse action predictions. We further integrate DT3 with the diffusion model using a unified optimization objective. With experiments on multiple tasks of Gym and AntMaze in the D4RL benchmark, our DT3 model without diffusion refinement demonstrates improved performance over standard DT, while DRDT3 further achieves superior results compared to state-of-the-art conventional offline RL and DT-based methods.
2501.06719
Hierarchical Sampling-based Planner with LTL Constraints and Text Prompting
cs.RO cs.SY eess.SY
This project introduces a hierarchical planner integrating Linear Temporal Logic (LTL) constraints with natural language prompting for robot motion planning. The framework decomposes maps into regions, generates directed graphs, and converts them into transition systems for high-level planning. Text instructions are translated into LTL formulas and converted to Deterministic Finite Automata (DFA) for sequential goal-reaching tasks while adhering to safety constraints. High-level plans, derived via Breadth-First Search (BFS), guide low-level planners like Exploring Random Trees (RRT) and Probabilistic Roadmaps (PRM) for obstacle-avoidant navigation along with LTL tasks. The approach demonstrates adaptability to various task complexities, though challenges such as graph construction overhead and suboptimal path generation remain. Future directions include extending to considering terrain conditions and incorporating higher-order dynamics.