id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2502.03891
Counterfactual Query Rewriting to Use Historical Relevance Feedback
cs.IR
When a retrieval system receives a query it has encountered before, previous relevance feedback, such as clicks or explicit judgments can help to improve retrieval results. However, the content of a previously relevant document may have changed, or the document might not be available anymore. Despite this evolved corpus, we counterfactually use these previously relevant documents as relevance signals. In this paper we proposed approaches to rewrite user queries and compare them against a system that directly uses the previous qrels for the ranking. We expand queries with terms extracted from the previously relevant documents or derive so-called keyqueries that rank the previously relevant documents to the top of the current corpus. Our evaluation in the CLEF LongEval scenario shows that rewriting queries with historical relevance feedback improves the retrieval effectiveness and even outperforms computationally expensive transformer-based approaches.
2502.03895
Rule-Based Modeling of Low-Dimensional Data with PCA and Binary Particle Swarm Optimization (BPSO) in ANFIS
cs.CV
Fuzzy rule-based systems interpret data in low-dimensional domains, providing transparency and interpretability. In contrast, deep learning excels in complex tasks like image and speech recognition but is prone to overfitting in sparse, unstructured, or low-dimensional data. This interpretability is crucial in fields like healthcare and finance. Traditional rule-based systems, especially ANFIS with grid partitioning, suffer from exponential rule growth as dimensionality increases. We propose a strategic rule-reduction model that applies Principal Component Analysis (PCA) on normalized firing strengths to obtain linearly uncorrelated components. Binary Particle Swarm Optimization (BPSO) selectively refines these components, significantly reducing the number of rules while preserving precision in decision-making. A custom parameter update mechanism fine-tunes specific ANFIS layers by dynamically adjusting BPSO parameters, avoiding local minima. We validated our approach on standard UCI respiratory, keel classification, regression datasets, and a real-world ischemic stroke dataset, demonstrating adaptability and practicality. Results indicate fewer rules, shorter training, and high accuracy, underscoring the methods effectiveness for low-dimensional interpretability and complex data scenarios. This synergy of fuzzy logic and optimization fosters robust solutions. Our method contributes a powerful framework for interpretable AI in multiple domains. It addresses dimensionality, ensuring a rule base.
2502.03897
UniForm: A Unified Diffusion Transformer for Audio-Video Generation
cs.MM cs.AI cs.CV cs.SD eess.AS
As a natural multimodal content, audible video delivers an immersive sensory experience. Consequently, audio-video generation systems have substantial potential. However, existing diffusion-based studies mainly employ relatively independent modules for generating each modality, which lack exploration of shared-weight generative modules. This approach may under-use the intrinsic correlations between audio and visual modalities, potentially resulting in sub-optimal generation quality. To address this, we propose UniForm, a unified diffusion transformer designed to enhance cross-modal consistency. By concatenating auditory and visual information, UniForm learns to generate audio and video simultaneously within a unified latent space, facilitating the creation of high-quality and well-aligned audio-visual pairs. Extensive experiments demonstrate the superior performance of our method in joint audio-video generation, audio-guided video generation, and video-guided audio generation tasks. Our demos are available at https://uniform-t2av.github.io/.
2502.03900
How to introduce an initial crack in phase field simulations to accurately predict the linear elastic fracture propagation threshold?
cs.CE
Variational phase field fracture models are now widely used to simulate crack propagation in structures. A critical aspect of these simulations is the correct determination of the propagation threshold of pre-existing cracks, as it highly relies on how the initial cracks are implemented. While prior studies briefly discuss initial crack implementation techniques, we present here a systematic investigation. Various techniques to introduce initial cracks in phase field fracture simulations are tested, from the crack explicit meshing to the replacement by a fully damaged phase field, including different variants for the boundary conditions. Our focus here is on phase field models aiming to approximate, in the $\Gamma$-convergence limit, Griffith quasi-static propagation in the framework of Linear Elastic Fracture Mechanics. Therefore, a sharp crack model from classic linear elastic fracture mechanics based on Griffith criterion is the reference in this work. To assess the different techniques to introduce initial cracks, we rely on path-following methods to compute the sharp crack and the phase field smeared crack solutions. The underlying idea is that path-following ensures staying at equilibrium at each instant so that any difference between phase field and sharp crack models can be attributed to numerical artifacts. Thus, by comparing the results from both models, we can provide practical recommendations for reliably incorporating initial cracks in phase field fracture simulations. The comparison shows that an improper initial crack implementation often requires the smeared crack to transition to a one-element-wide phase band to adequately represent a displacement jump along a crack. This transition increases the energy required to propagate the crack, leading to a significant overshoot in the force-displacement response. The take-home message is that to predict the propagation threshold accurately and avoid artificial toughening; the crack must be initialized either setting the phase field to its damage state over a one-element-wide band or meshing the crack explicitly as a one-element-wide slit and imposing the fully cracked state on the crack surface.
2502.03901
LeAP: Consistent multi-domain 3D labeling using Foundation Models
cs.CV cs.RO
Availability of datasets is a strong driver for research on 3D semantic understanding, and whilst obtaining unlabeled 3D point cloud data is straightforward, manually annotating this data with semantic labels is time-consuming and costly. Recently, Vision Foundation Models (VFMs) enable open-set semantic segmentation on camera images, potentially aiding automatic labeling. However,VFMs for 3D data have been limited to adaptations of 2D models, which can introduce inconsistencies to 3D labels. This work introduces Label Any Pointcloud (LeAP), leveraging 2D VFMs to automatically label 3D data with any set of classes in any kind of application whilst ensuring label consistency. Using a Bayesian update, point labels are combined into voxels to improve spatio-temporal consistency. A novel 3D Consistency Network (3D-CN) exploits 3D information to further improve label quality. Through various experiments, we show that our method can generate high-quality 3D semantic labels across diverse fields without any manual labeling. Further, models adapted to new domains using our labels show up to a 34.2 mIoU increase in semantic segmentation tasks.
2502.03902
Geometric Stabilization of Virtual Nonlinear Nonholonomic Constraints
eess.SY cs.SY math.DS math.OC
In this paper, we address the problem of stabilizing a system around a desired manifold determined by virtual nonlinear nonholonomic constraints. Virtual constraints are relationships imposed on a control system that are rendered invariant through feedback control. Virtual nonholonomic constraints represent a specific class of virtual constraints that depend on the system's velocities in addition to its configurations. We derive a control law under which a mechanical control system achieves exponential convergence to the virtual constraint submanifold, and rendering it control-invariant. The proposed controller's performance is validated through simulation results in two distinct applications: flocking motion in multi-agent systems and the control of an unmanned surface vehicle (USV) navigating a stream.
2502.03904
A Gaussian-Sinc Pulse Shaping Filter for Zak-OTFS
cs.IT eess.SP math.IT
The choice of delay-Doppler domain (DD) pulse shaping filter plays an important role in determining the performance of Zak-OTFS. Sinc filter has good main lobe characteristics (with nulls at information grid points) which is good for equalization/detection, but has high side lobes which are detrimental for input-output (I/O) relation estimation. Whereas, Gaussian filter is highly localized with very low side lobes which is good for I/O relation estimation, but has poor main lobe characteristics which is not good for equalization/detection. In this paper, we propose a new filter, termed as {\em Gaussian-sinc (GS) filter}, which inherits the complementary strengths of both Gaussian and sinc filters. The proposed filter does not incur time or bandwidth expansion. We derive closed-form expressions for the I/O relation and noise covariance of Zak-OTFS with the proposed GS filter. We evaluate the Zak-OTFS performance for different pulse shaping filters with I/O relation estimated using exclusive and embedded pilots. Our results show that the proposed GS filter achieves better bit error rate (BER) performance compared to other filters reported in the literature. For example, with model-free I/O relation estimation using embedded pilot and 8-QAM, the proposed GS filter achieves an SNR gain of about 4 dB at $10^{-2}$ uncoded BER compared to Gaussian and sinc filters, and the SNR gain becomes more than 6 dB at a coded BER of $10^{-4}$ with rate-1/2 coding.
2502.03907
No Free Lunch in Annotation either: An objective evaluation of foundation models for streamlining annotation in animal tracking
cs.CV
We analyze the capabilities of foundation models addressing the tedious task of generating annotations for animal tracking. Annotating a large amount of data is vital and can be a make-or-break factor for the robustness of a tracking model. Robustness is particularly crucial in animal tracking, as accurate tracking over long time horizons is essential for capturing the behavior of animals. However, generating additional annotations using foundation models can be counterproductive, as the quality of the annotations is just as important. Poorly annotated data can introduce noise and inaccuracies, ultimately compromising the performance and accuracy of the trained model. Over-reliance on automated annotations without ensuring precision can lead to diminished results, making careful oversight and quality control essential in the annotation process. Ultimately, we demonstrate that a thoughtful combination of automated annotations and manually annotated data is a valuable strategy, yielding an IDF1 score of 80.8 against blind usage of SAM2 video with an IDF1 score of 65.6.
2502.03909
Technical Report: Generating the WEB-IDS23 Dataset
cs.NI cs.CR cs.LG
Anomaly-based Network Intrusion Detection Systems (NIDS) require correctly labelled, representative and diverse datasets for an accurate evaluation and development. However, several widely used datasets do not include labels which are fine-grained enough and, together with small sample sizes, can lead to overfitting issues that also remain undetected when using test data. Additionally, the cybersecurity sector is evolving fast, and new attack mechanisms require the continuous creation of up-to-date datasets. To address these limitations, we developed a modular traffic generator that can simulate a wide variety of benign and malicious traffic. It incorporates multiple protocols, variability through randomization techniques and can produce attacks along corresponding benign traffic, as it occurs in real-world scenarios. Using the traffic generator, we create a dataset capturing over 12 million samples with 82 flow-level features and 21 fine-grained labels. Additionally, we include several web attack types which are often underrepresented in other datasets.
2502.03914
A Flexible FBG-Based Contact Force Sensor for Robotic Gripping Systems
cs.RO cs.SY eess.SY
Soft robotic grippers demonstrate great potential for gently and safely handling objects; however, their full potential for executing precise and secure grasping has been limited by the lack of integrated sensors, leading to problems such as slippage and excessive force exertion. To address this challenge, we present a small and highly sensitive Fiber Bragg Grating-based force sensor designed for accurate contact force measurement. The flexible force sensor comprises a 3D-printed TPU casing with a small bump and uvula structure, a dual FBG array, and a protective tube. A series of tests have been conducted to evaluate the effectiveness of the proposed force sensor, including force calibration, repeatability test, hysteresis study, force measurement comparison, and temperature calibration and compensation tests. The results demonstrated good repeatability, with a force measurement range of 4.69 N, a high sensitivity of approximately 1169.04 pm/N, a root mean square error (RMSE) of 0.12 N, and a maximum hysteresis of 4.83%. When compared to a commercial load cell, the sensor showed a percentage error of 2.56% and an RMSE of 0.14 N. Besides, the proposed sensor validated its temperature compensation effectiveness, with a force RMSE of 0.01 N over a temperature change of 11 Celsius degree. The sensor was integrated with a soft grow-and-twine gripper to monitor interaction forces between different objects and the robotic gripper. Closed-loop force control was applied during automated pick-and-place tasks and significantly improved gripping stability, as demonstrated in tests. This force sensor can be used across manufacturing, agriculture, healthcare (like prosthetic hands), logistics, and packaging, to provide situation awareness and higher operational efficiency.
2502.03916
Experiments with Large Language Models on Retrieval-Augmented Generation for Closed-Source Simulation Software
cs.CL cs.AI
Large Language Models (LLMs) are increasingly helpful in text generation, even writing code in programming languages based on user prompts written in natural language. They are even applied to generate simulation models for multibody systems from natural language. Research results suggest that LLMs surpass the mere replication of existing code examples, where some LLMs have been trained on an open-source multibody simulation code. However, for closed-source simulation software, such results are not to be expected as their ideas and concepts might differ from other publicly available ones. LLMs can hallucinate for knowledge-intensive tasks, such as model creation, which can lead to wrong responses. This is especially the case for the LLM unknown closed-source simulation software. The same applies to other internal knowledge kept private to protect intellectual property or data privacy. The Retrieval-Augmented Generation (RAG) approach might yield a solution for these knowledge-intensive tasks. This paper explores the application of RAG to closed-source simulation software and presents first experiments. After a brief introduction to LLMs, the RAG approach, and the simulation method applied by the close-source simulation software, several examples are provided to test LLMs' knowledge of the simulation software and the creation of simulation models using two RAG systems. The examples show promising results indicating the benefits of applying RAG systems to closed-source simulation software, helping to access their knowledge. Nevertheless, they also reveal gaps in the applied information and open questions for further research.
2502.03917
On the existence of strong functional observer
eess.SY cs.SY
For arbitrary linear time-invariant systems, the existence of a strong functional observer is investigated. Such observer determines, from the available measurement on the plant, an estimate of a function of the state and the input. This estimate converges irrespective to initial state and input. This formulation encompass the cases of observer existence for known or unknown inputs and generalizes state-of-art. Necessary and sufficient conditions for such an existence are proposed, in the framework of state-space representation. These conditions are based on functional detectability property and its generalizations for arbitrary input, which include considerations on convergence of the estimation, irrespective to the initial state and the input. Known results on state detectability, input reconstruction or functional detectability are retrieved by particularizing the proposed conditions.
2502.03918
Adaptation of Task Goal States from Prior Knowledge
cs.RO cs.AI
This paper presents a framework to define a task with freedom and variability in its goal state. A robot could use this to observe the execution of a task and target a different goal from the observed one; a goal that is still compatible with the task description but would be easier for the robot to execute. We define the model of an environment state and an environment variation, and present experiments on how to interactively create the variation from a single task demonstration and how to use this variation to create an execution plan for bringing any environment into the goal state.
2502.03919
Blackwell's Approachability with Approximation Algorithms
math.OC cs.LG
We revisit Blackwell's celebrated approachability problem which considers a repeated vector-valued game between a player and an adversary. Motivated by settings in which the action set of the player or adversary (or both) is difficult to optimize over, for instance when it corresponds to the set of all possible solutions to some NP-Hard optimization problem, we ask what can the player guarantee \textit{efficiently}, when only having access to these sets via approximation algorithms with ratios $\alpha_{\mX} \geq 1$ and $ 1 \geq \alpha_{\mY} > 0$, respectively. Assuming the player has monotone preferences, in the sense that he does not prefer a vector-valued loss $\ell_1$ over $\ell_2$ if $\ell_2 \leq \ell_1$, we establish that given a Blackwell instance with an approachable target set $S$, the downward closure of the appropriately-scaled set $\alpha_{\mX}\alpha_{\mY}^{-1}S$ is \textit{efficiently} approachable with optimal rate. In case only the player's or adversary's set is equipped with an approximation algorithm, we give simpler and more efficient algorithms.
2502.03930
DiTAR: Diffusion Transformer Autoregressive Modeling for Speech Generation
eess.AS cs.AI cs.CL cs.LG cs.SD
Several recent studies have attempted to autoregressively generate continuous speech representations without discrete speech tokens by combining diffusion and autoregressive models, yet they often face challenges with excessive computational loads or suboptimal outcomes. In this work, we propose Diffusion Transformer Autoregressive Modeling (DiTAR), a patch-based autoregressive framework combining a language model with a diffusion transformer. This approach significantly enhances the efficacy of autoregressive models for continuous tokens and reduces computational demands. DiTAR utilizes a divide-and-conquer strategy for patch generation, where the language model processes aggregated patch embeddings and the diffusion transformer subsequently generates the next patch based on the output of the language model. For inference, we propose defining temperature as the time point of introducing noise during the reverse diffusion ODE to balance diversity and determinism. We also show in the extensive scaling analysis that DiTAR has superb scalability. In zero-shot speech generation, DiTAR achieves state-of-the-art performance in robustness, speaker similarity, and naturalness.
2502.03933
HEP-JEPA: A foundation model for collider physics using joint embedding predictive architecture
cs.LG hep-ex hep-ph
We present a transformer architecture-based foundation model for tasks at high-energy particle colliders such as the Large Hadron Collider. We train the model to classify jets using a self-supervised strategy inspired by the Joint Embedding Predictive Architecture. We use the JetClass dataset containing 100M jets of various known particles to pre-train the model with a data-centric approach -- the model uses a fraction of the jet constituents as the context to predict the embeddings of the unseen target constituents. Our pre-trained model fares well with other datasets for standard classification benchmark tasks. We test our model on two additional downstream tasks: top tagging and differentiating light-quark jets from gluon jets. We also evaluate our model with task-specific metrics and baselines and compare it with state-of-the-art models in high-energy physics. Project site: https://hep-jepa.github.io/
2502.03935
Thermal Model Calibration of a Squirrel-Cage Induction Machine
cs.CE
Accurate and efficient thermal simulations of induction machines are indispensable for detecting thermal hot spots and hence avoiding potential material failure in an early design stage. A goal is the better utilization of the machines with reduced safety margins due to a better knowledge of the critical conditions. In this work, the parameters of a two-dimensional induction machine model are calibrated according to evidence from measurements, by solving an inverse field problem. The set of parameters comprise material parameters as well as parameters that model three-dimensional effects. This allows a consideration of physical effects without explicit knowledge of its quantities. First, the accuracy of the approach is studied using an academic example in combination with synthetic data. Afterwards, it is successfully applied to a realistic induction machine model.
2502.03937
Quantifying Correlations of Machine Learning Models
cs.LG cs.NA math.NA
Machine Learning models are being extensively used in safety critical applications where errors from these models could cause harm to the user. Such risks are amplified when multiple machine learning models, which are deployed concurrently, interact and make errors simultaneously. This paper explores three scenarios where error correlations between multiple models arise, resulting in such aggregated risks. Using real-world data, we simulate these scenarios and quantify the correlations in errors of different models. Our findings indicate that aggregated risks are substantial, particularly when models share similar algorithms, training datasets, or foundational models. Overall, we observe that correlations across models are pervasive and likely to intensify with increased reliance on foundational models and widely used public datasets, highlighting the need for effective mitigation strategies to address these challenges.
2502.03938
Unravelling Causal Genetic Biomarkers of Alzheimer's Disease via Neuron to Gene-token Backtracking in Neural Architecture: A Groundbreaking Reverse-Gene-Finder Approach
cs.LG
Alzheimer's Disease (AD) affects over 55 million people globally, yet the key genetic contributors remain poorly understood. Leveraging recent advancements in genomic foundation models, we present the innovative Reverse-Gene-Finder technology, a ground-breaking neuron-to-gene-token backtracking approach in a neural network architecture to elucidate the novel causal genetic biomarkers driving AD onset. Reverse-Gene-Finder comprises three key innovations. Firstly, we exploit the observation that genes with the highest probability of causing AD, defined as the most causal genes (MCGs), must have the highest probability of activating those neurons with the highest probability of causing AD, defined as the most causal neurons (MCNs). Secondly, we utilize a gene token representation at the input layer to allow each gene (known or novel to AD) to be represented as a discrete and unique entity in the input space. Lastly, in contrast to the existing neural network architectures, which track neuron activations from the input layer to the output layer in a feed-forward manner, we develop an innovative backtracking method to track backwards from the MCNs to the input layer, identifying the Most Causal Tokens (MCTs) and the corresponding MCGs. Reverse-Gene-Finder is highly interpretable, generalizable, and adaptable, providing a promising avenue for application in other disease scenarios.
2502.03943
Multimodal Data-Driven Classification of Mental Disorders: A Comprehensive Approach to Diagnosing Depression, Anxiety, and Schizophrenia
cs.LG
This study investigates the potential of multimodal data integration, which combines electroencephalogram (EEG) data with sociodemographic characteristics like age, sex, education, and intelligence quotient (IQ), to diagnose mental diseases like schizophrenia, depression, and anxiety. Using Apache Spark and convolutional neural networks (CNNs), a data-driven classification pipeline has been developed for big data environment to effectively analyze massive datasets. In order to evaluate brain activity and connection patterns associated with mental disorders, EEG parameters such as power spectral density (PSD) and coherence are examined. The importance of coherence features is highlighted by comparative analysis, which shows significant improvement in classification accuracy and robustness. This study emphasizes the significance of holistic approaches for efficient diagnostic tools by integrating a variety of data sources. The findings open the door for creative, data-driven approaches to treating psychiatric diseases by demonstrating the potential of utilizing big data, sophisticated deep learning methods, and multimodal datasets to enhance the precision, usability, and comprehension of mental health diagnostics.
2502.03944
Exact Covariance Characterization for Controlled Linear Systems subject to Stochastic Parametric and Additive Uncertainties
eess.SY cs.SY
This work addresses the exact characterization of the covariance dynamics related to linear discrete-time systems subject to both additive and parametric stochastic uncertainties that are potentially unbounded. The derived exact representation allows to understand how the covariance of the multiplicative parametric uncertainties affects the stability of the state covariance dynamics through a transformation of the parameters covariance matrix, allowing therefore to address the problem of control design for state covariance dynamics in this context. Numerical results assess this new characterization by comparing it to the empirical covariance and illustrating the control design problem.
2502.03945
Afrispeech-Dialog: A Benchmark Dataset for Spontaneous English Conversations in Healthcare and Beyond
cs.CL
Speech technologies are transforming interactions across various sectors, from healthcare to call centers and robots, yet their performance on African-accented conversations remains underexplored. We introduce Afrispeech-Dialog, a benchmark dataset of 50 simulated medical and non-medical African-accented English conversations, designed to evaluate automatic speech recognition (ASR) and related technologies. We assess state-of-the-art (SOTA) speaker diarization and ASR systems on long-form, accented speech, comparing their performance with native accents and discover a 10%+ performance degradation. Additionally, we explore medical conversation summarization capabilities of large language models (LLMs) to demonstrate the impact of ASR errors on downstream medical summaries, providing insights into the challenges and opportunities for speech technologies in the Global South. Our work highlights the need for more inclusive datasets to advance conversational AI in low-resource settings.
2502.03946
CleanSurvival: Automated data preprocessing for time-to-event models using reinforcement learning
cs.LG
Data preprocessing is a critical yet frequently neglected aspect of machine learning, often paid little attention despite its potentially significant impact on model performance. While automated machine learning pipelines are starting to recognize and integrate data preprocessing into their solutions for classification and regression tasks, this integration is lacking for more specialized tasks like survival or time-to-event models. As a result, survival analysis not only faces the general challenges of data preprocessing but also suffers from the lack of tailored, automated solutions in this area. To address this gap, this paper presents 'CleanSurvival', a reinforcement-learning-based solution for optimizing preprocessing pipelines, extended specifically for survival analysis. The framework can handle continuous and categorical variables, using Q-learning to select which combination of data imputation, outlier detection and feature extraction techniques achieves optimal performance for a Cox, random forest, neural network or user-supplied time-to-event model. The package is available on GitHub: https://github.com/datasciapps/CleanSurvival Experimental benchmarks on real-world datasets show that the Q-learning-based data preprocessing results in superior predictive performance to standard approaches, finding such a model up to 10 times faster than undirected random grid search. Furthermore, a simulation study demonstrates the effectiveness in different types and levels of missingness and noise in the data.
2502.03948
Enhancing Online Learning Efficiency Through Heterogeneous Resource Integration with a Multi-Agent RAG System
cs.AI cs.CL cs.MA
Efficient online learning requires seamless access to diverse resources such as videos, code repositories, documentation, and general web content. This poster paper introduces early-stage work on a Multi-Agent Retrieval-Augmented Generation (RAG) System designed to enhance learning efficiency by integrating these heterogeneous resources. Using specialized agents tailored for specific resource types (e.g., YouTube tutorials, GitHub repositories, documentation websites, and search engines), the system automates the retrieval and synthesis of relevant information. By streamlining the process of finding and combining knowledge, this approach reduces manual effort and enhances the learning experience. A preliminary user study confirmed the system's strong usability and moderate-high utility, demonstrating its potential to improve the efficiency of knowledge acquisition.
2502.03950
LR0.FM: Low-Resolution Zero-shot Classification Benchmark For Foundation Models
cs.CV
Visual-language foundation Models (FMs) exhibit remarkable zero-shot generalization across diverse tasks, largely attributed to extensive pre-training on largescale datasets. However, their robustness on low-resolution/pixelated (LR) images, a common challenge in real-world scenarios, remains underexplored. We introduce LR0.FM, a comprehensive benchmark evaluating the impact of low resolution on the zero-shot classification performance of 10 FM(s) across 66 backbones and 15 datasets. We propose a novel metric, Weighted Aggregated Robustness, to address the limitations of existing metrics and better evaluate model performance across resolutions and datasets. Our key findings show that: (i) model size positively correlates with robustness to resolution degradation, (ii) pre-training dataset quality is more important than its size, and (iii) fine-tuned and higher resolution models are less robust against LR. Our analysis further reveals that the model makes semantically reasonable predictions at LR, and the lack of fine-grained details in input adversely impacts the model's initial layers more than the deeper layers. We use these insights and introduce a simple strategy, LR-TK0, to enhance the robustness of models without compromising their pre-trained weights. We demonstrate the effectiveness of LR-TK0 for robustness against low-resolution across several datasets and its generalization capability across backbones and other approaches. Code is available at https://github.com/shyammarjit/LR0.FM
2502.03952
Bridging the inference gap in Mutimodal Variational Autoencoders
cs.LG stat.ML
From medical diagnosis to autonomous vehicles, critical applications rely on the integration of multiple heterogeneous data modalities. Multimodal Variational Autoencoders offer versatile and scalable methods for generating unobserved modalities from observed ones. Recent models using mixturesof-experts aggregation suffer from theoretically grounded limitations that restrict their generation quality on complex datasets. In this article, we propose a novel interpretable model able to learn both joint and conditional distributions without introducing mixture aggregation. Our model follows a multistage training process: first modeling the joint distribution with variational inference and then modeling the conditional distributions with Normalizing Flows to better approximate true posteriors. Importantly, we also propose to extract and leverage the information shared between modalities to improve the conditional coherence of generated samples. Our method achieves state-of-the-art results on several benchmark datasets.
2502.03953
Fairness Aware Reinforcement Learning via Proximal Policy Optimization
cs.MA cs.LG
Fairness in multi-agent systems (MAS) focuses on equitable reward distribution among agents in scenarios involving sensitive attributes such as race, gender, or socioeconomic status. This paper introduces fairness in Proximal Policy Optimization (PPO) with a penalty term derived from demographic parity, counterfactual fairness, and conditional statistical parity. The proposed method balances reward maximisation with fairness by integrating two penalty components: a retrospective component that minimises disparities in past outcomes and a prospective component that ensures fairness in future decision-making. We evaluate our approach in the Allelopathic Harvest game, a cooperative and competitive MAS focused on resource collection, where some agents possess a sensitive attribute. Experiments demonstrate that fair-PPO achieves fairer policies across all fairness metrics than classic PPO. Fairness comes at the cost of reduced rewards, namely the Price of Fairness, although agents with and without the sensitive attribute renounce comparable amounts of rewards. Additionally, the retrospective and prospective penalties effectively change the agents' behaviour and improve fairness. These findings underscore the potential of fair-PPO to address fairness challenges in MAS.
2502.03954
MAQInstruct: Instruction-based Unified Event Relation Extraction
cs.CL cs.AI
Extracting event relations that deviate from known schemas has proven challenging for previous methods based on multi-class classification, MASK prediction, or prototype matching. Recent advancements in large language models have shown impressive performance through instruction tuning. Nevertheless, in the task of event relation extraction, instruction-based methods face several challenges: there are a vast number of inference samples, and the relations between events are non-sequential. To tackle these challenges, we present an improved instruction-based event relation extraction framework named MAQInstruct. Firstly, we transform the task from extracting event relations using given event-event instructions to selecting events using given event-relation instructions, which reduces the number of samples required for inference. Then, by incorporating a bipartite matching loss, we reduce the dependency of the instruction-based method on the generation sequence. Our experimental results demonstrate that MAQInstruct significantly improves the performance of event relation extraction across multiple LLMs.
2502.03957
Improving the Perturbation-Based Explanation of Deepfake Detectors Through the Use of Adversarially-Generated Samples
cs.CV cs.AI cs.CR
In this paper, we introduce the idea of using adversarially-generated samples of the input images that were classified as deepfakes by a detector, to form perturbation masks for inferring the importance of different input features and produce visual explanations. We generate these samples based on Natural Evolution Strategies, aiming to flip the original deepfake detector's decision and classify these samples as real. We apply this idea to four perturbation-based explanation methods (LIME, SHAP, SOBOL and RISE) and evaluate the performance of the resulting modified methods using a SOTA deepfake detection model, a benchmarking dataset (FaceForensics++) and a corresponding explanation evaluation framework. Our quantitative assessments document the mostly positive contribution of the proposed perturbation approach in the performance of explanation methods. Our qualitative analysis shows the capacity of the modified explanation methods to demarcate the manipulated image regions more accurately, and thus to provide more useful explanations.
2502.03958
Non-convex composite federated learning with heterogeneous data
cs.LG cs.DC
We propose an innovative algorithm for non-convex composite federated learning that decouples the proximal operator evaluation and the communication between server and clients. Moreover, each client uses local updates to communicate less frequently with the server, sends only a single d-dimensional vector per communication round, and overcomes issues with client drift. In the analysis, challenges arise from the use of decoupling strategies and local updates in the algorithm, as well as from the non-convex and non-smooth nature of the problem. We establish sublinear and linear convergence to a bounded residual error under general non-convexity and the proximal Polyak-Lojasiewicz inequality, respectively. In the numerical experiments, we demonstrate the superiority of our algorithm over state-of-the-art methods on both synthetic and real datasets.
2502.03960
Bilevel Multi-Armed Bandit-Based Hierarchical Reinforcement Learning for Interaction-Aware Self-Driving at Unsignalized Intersections
cs.RO
In this work, we present BiM-ACPPO, a bilevel multi-armed bandit-based hierarchical reinforcement learning framework for interaction-aware decision-making and planning at unsignalized intersections. Essentially, it proactively takes the uncertainties associated with surrounding vehicles (SVs) into consideration, which encompass those stemming from the driver's intention, interactive behaviors, and the varying number of SVs. Intermediate decision variables are introduced to enable the high-level RL policy to provide an interaction-aware reference, for guiding low-level model predictive control (MPC) and further enhancing the generalization ability of the proposed framework. By leveraging the structured nature of self-driving at unsignalized intersections, the training problem of the RL policy is modeled as a bilevel curriculum learning task, which is addressed by the proposed Exp3.S-based BiMAB algorithm. It is noteworthy that the training curricula are dynamically adjusted, thereby facilitating the sample efficiency of the RL training process. Comparative experiments are conducted in the high-fidelity CARLA simulator, and the results indicate that our approach achieves superior performance compared to all baseline methods. Furthermore, experimental results in two new urban driving scenarios clearly demonstrate the commendable generalization performance of the proposed method.
2502.03962
Quantum Circuit Design using a Progressive Widening Monte Carlo Tree Search
quant-ph cs.AI cs.ET
The performance of Variational Quantum Algorithms (VQAs) strongly depends on the choice of the parameterized quantum circuit to optimize. One of the biggest challenges in VQAs is designing quantum circuits tailored to the particular problem and to the quantum hardware. This article proposes a gradient-free Monte Carlo Tree Search (MCTS) technique to automate the process of quantum circuit design. It introduces a novel formulation of the action space based on a sampling scheme and a progressive widening technique to explore the space dynamically. When testing our MCTS approach on the domain of random quantum circuits, MCTS approximates unstructured circuits under different values of stabilizer R\'enyi entropy. It turns out that MCTS manages to approximate the benchmark quantum states independently from their degree of nonstabilizerness. Next, our technique exhibits robustness across various application domains, including quantum chemistry and systems of linear equations. Compared to previous MCTS research, our technique reduces the number of quantum circuit evaluations by a factor of 10 to 100 while achieving equal or better results. In addition, the resulting quantum circuits have up to three times fewer CNOT gates.
2502.03963
AL-PINN: Active Learning-Driven Physics-Informed Neural Networks for Efficient Sample Selection in Solving Partial Differential Equations
cs.LG
Physics-Informed Neural Networks (PINNs) have emerged as a promising approach for solving Partial Differential Equations (PDEs) by incorporating physical constraints into deep learning models. However, standard PINNs often require a large number of training samples to achieve high accuracy, leading to increased computational costs. To address this issue, we propose Active Learning-Driven PINNs (AL-PINN), which integrates Uncertainty Quantification (UQ) and Active Learning (AL) strategies to optimize sample selection dynamically. AL-PINN utilizes Monte Carlo Dropout to estimate epistemic uncertainty in the model predictions, enabling the adaptive selection of high-uncertainty regions for additional training. This approach significantly enhances learning efficiency by focusing computational resources on the most informative data points. We evaluate AL-PINN on benchmark PDE problems with known analytical solutions and real-world WeatherBench climate data. Our results demonstrate that AL-PINN achieves comparable or superior accuracy compared to traditional PINNs while reducing the number of required training samples. The proposed framework is particularly beneficial for scientific and engineering applications where data collection is expensive or limited, such as climate modeling, medical simulations, and material science. Our findings highlight the potential of active learning in accelerating PINN-based PDE solvers while maintaining high accuracy and computational efficiency.
2502.03965
Innovative Framework for Early Estimation of Mental Disorder Scores to Enable Timely Interventions
cs.LG
Individual's general well-being is greatly impacted by mental health conditions including depression and Post-Traumatic Stress Disorder (PTSD), underscoring the importance of early detection and precise diagnosis in order to facilitate prompt clinical intervention. An advanced multimodal deep learning system for the automated classification of PTSD and depression is presented in this paper. Utilizing textual and audio data from clinical interview datasets, the method combines features taken from both modalities by combining the architectures of LSTM (Long Short Term Memory) and BiLSTM (Bidirectional Long Short-Term Memory).Although text features focus on speech's semantic and grammatical components; audio features capture vocal traits including rhythm, tone, and pitch. This combination of modalities enhances the model's capacity to identify minute patterns connected to mental health conditions. Using test datasets, the proposed method achieves classification accuracies of 92% for depression and 93% for PTSD, outperforming traditional unimodal approaches and demonstrating its accuracy and robustness.
2502.03966
MultiFloodSynth: Multi-Annotated Flood Synthetic Dataset Generation
cs.CV cs.AI cs.LG
In this paper, we present synthetic data generation framework for flood hazard detection system. For high fidelity and quality, we characterize several real-world properties into virtual world and simulate the flood situation by controlling them. For the sake of efficiency, recent generative models in image-to-3D and urban city synthesis are leveraged to easily composite flood environments so that we avoid data bias due to the hand-crafted manner. Based on our framework, we build the flood synthetic dataset with 5 levels, dubbed MultiFloodSynth which contains rich annotation types like normal map, segmentation, 3D bounding box for a variety of downstream task. In experiments, our dataset demonstrate the enhanced performance of flood hazard detection with on-par realism compared with real dataset.
2502.03971
RWKV-UI: UI Understanding with Enhanced Perception and Reasoning
cs.CV cs.HC
Existing Visual Language Modelsoften struggle with information loss and limited reasoning abilities when handling high-resolution web interfaces that combine complex visual, textual, and interactive elements. These challenges are particularly evident in tasks requiring webpage layout comprehension and multi-step interactive reasoning. To address these challenges, we propose RWKV-UI, a Visual Language Model based on the RWKV architecture, specifically designed to handle high-resolution UI images. During model training, we introduce layout detection as a visual prompt to help the model better understand the webpage layout structures. Additionally, we design a visual prompt based on the Chain-of-Thought(CoT) mechanism, which enhances the model's ability to understand and reason about webpage content through reasoning chains. Experimental results show that RWKV-UI demonstrates significant performance improvements in high-resolution UI understanding and interactive reasoning tasks.
2502.03974
Spatiotemporal Trajectory Tracking Method for Vehicles Incorporating Lead-Lag Judgement
eess.SY cs.SY
In the domain of intelligent transportation systems, especially within the context of autonomous vehicle control, the preemptive holistic collaborative system has been presented as a promising solution to bring a remarkable enhancement in traffic efficiency and a substantial reduction in the accident rate, demonstrating a great potential of development. In order to ensure this system operates as intended, accurate tracking of the spatiotemporal trajectory is of crucial significance. Moreover, minimizing the tracking error is a necessary step in this process. To this end, a novel lead-lag judgment mechanism is proposed. This mechanism precisely quantifies the longitudinal positional deviation between the vehicle and the target trajectory over time, then the deviation is corrected with a real - time acceleration compensation strategy, as a result, the accuracy and reliability of trajectory tracking are significantly enhanced. Real - vehicle experiments were conducted in a dedicated test field to validate the feasibility of this innovative approach empirically. Subsequently, the obtained tracking data was subsequent processed using the lead-lag judgment mechanism. In this step, we carefully analyzed the spatiotemporal error patterns between the vehicle and the target trajectory under different alignments and speeds. Finally, using real highway speed and alignment data, we conducted comprehensive spatiotemporal trajectory tracking simulations. Through experiments and simulations, tracking errors maintained in an acceptable range and reasonable spatiotemporal distance is given during the preemptive merging process on highway ramps. Overall, this study offers valuable insights for highway ramp emerging safety. Future work can expand on these findings.
2502.03976
Small Signal Stability Analysis of Kurdistan Regional Power System
eess.SY cs.SY
This paper presents for the first time a mathematical model for evaluating the Planned Kurdistan Regional Power System (KRPS) for its ability to maintain stability under small disturbances and fluctuations during normal operating conditions. To achieve this objective, practical field data, manufacture's datasheets, related IEEE task force reports have been used to build a complete mathematical model in MATLAB/SIMULINK/SimPowerSystem environment. New modules have been established and added to the platform wherever it does not support special type of elements. The model represents accurately all the power system components involved in physical phenomena of system dynamic oscillations. The model consists of 53 transmission lines, 35 nodes and 6 generating stations. The system is simulated under different configurations and settings; the dynamic behaviors associated with each configuration are recorded and analyzed accordingly.
2502.03979
Towards Unified Music Emotion Recognition across Dimensional and Categorical Models
cs.SD cs.AI eess.AS
One of the most significant challenges in Music Emotion Recognition (MER) comes from the fact that emotion labels can be heterogeneous across datasets with regard to the emotion representation, including categorical (e.g., happy, sad) versus dimensional labels (e.g., valence-arousal). In this paper, we present a unified multitask learning framework that combines these two types of labels and is thus able to be trained on multiple datasets. This framework uses an effective input representation that combines musical features (i.e., key and chords) and MERT embeddings. Moreover, knowledge distillation is employed to transfer the knowledge of teacher models trained on individual datasets to a student model, enhancing its ability to generalize across multiple tasks. To validate our proposed framework, we conducted extensive experiments on a variety of datasets, including MTG-Jamendo, DEAM, PMEmo, and EmoMusic. According to our experimental results, the inclusion of musical features, multitask learning, and knowledge distillation significantly enhances performance. In particular, our model outperforms the state-of-the-art models, including the best-performing model from the MediaEval 2021 competition on the MTG-Jamendo dataset. Our work makes a significant contribution to MER by allowing the combination of categorical and dimensional emotion labels in one unified framework, thus enabling training across datasets.
2502.03982
Temporal Distribution Shift in Real-World Pharmaceutical Data: Implications for Uncertainty Quantification in QSAR Models
cs.LG
The estimation of uncertainties associated with predictions from quantitative structure-activity relationship (QSAR) models can accelerate the drug discovery process by identifying promising experiments and allowing an efficient allocation of resources. Several computational tools exist that estimate the predictive uncertainty in machine learning models. However, deviations from the i.i.d. setting have been shown to impair the performance of these uncertainty quantification methods. We use a real-world pharmaceutical dataset to address the pressing need for a comprehensive, large-scale evaluation of uncertainty estimation methods in the context of realistic distribution shifts over time. We investigate the performance of several uncertainty estimation methods, including ensemble-based and Bayesian approaches. Furthermore, we use this real-world setting to systematically assess the distribution shifts in label and descriptor space and their impact on the capability of the uncertainty estimation methods. Our study reveals significant shifts over time in both label and descriptor space and a clear connection between the magnitude of the shift and the nature of the assay. Moreover, we show that pronounced distribution shifts impair the performance of popular uncertainty estimation methods used in QSAR models. This work highlights the challenges of identifying uncertainty quantification methods that remain reliable under distribution shifts introduced by real-world data.
2502.03984
PGB: One-Shot Pruning for BERT via Weight Grouping and Permutation
cs.CL cs.AI
Large pretrained language models such as BERT suffer from slow inference and high memory usage, due to their huge size. Recent approaches to compressing BERT rely on iterative pruning and knowledge distillation, which, however, are often too complicated and computationally intensive. This paper proposes a novel semi-structured one-shot pruning method for BERT, called $\textit{Permutation and Grouping for BERT}$ (PGB), which achieves high compression efficiency and sparsity while preserving accuracy. To this end, PGB identifies important groups of individual weights by permutation and prunes all other weights as a structure in both multi-head attention and feed-forward layers. Furthermore, if no important group is formed in a particular layer, PGB drops the entire layer to produce an even more compact model. Our experimental results on BERT$_{\text{BASE}}$ demonstrate that PGB outperforms the state-of-the-art structured pruning methods in terms of computational cost and accuracy preservation.
2502.03988
Tight Bounds on Jensen's Gap: Novel Approach with Applications in Generative Modeling
cs.LG
Among various mathematical tools of particular interest are those that provide a common basis for researchers in different scientific fields. One of them is Jensen's inequality, which states that the expectation of a convex function is greater than or equal to the function evaluated at the expectation. The resulting difference, known as Jensen's gap, became the subject of investigation by both the statistical and machine learning communities. Among many related topics, finding lower and upper bounds on Jensen's gap (under different assumptions on the underlying function and distribution) has recently become a problem of particular interest. In our paper, we take another step in this direction by providing a novel general and mathematically rigorous technique, motivated by the recent results of Struski et al. (2023). In addition, by studying in detail the case of the logarithmic function and the log-normal distribution, we explore a method for tightly estimating the log-likelihood of generative models trained on real-world datasets. Furthermore, we present both analytical and experimental arguments in support of the superiority of our approach in comparison to existing state-of-the-art solutions, contingent upon fulfillment of the criteria set forth by theoretical studies and corresponding experiments on synthetic data.
2502.03990
Frequency Control and Power Sharing in Combined Heat and Power Networks
eess.SY cs.SY
We consider the problem of using district heating systems as ancillary services for primary frequency control in power networks. We propose a novel power sharing scheme for heating systems based on the average temperature, which enables an optimal power allocation among the diverse heat sources without having a prior knowledge of the disturbances. We then discuss two approaches for heating systems to contribute to frequency regulation in power networks. We show that both approaches ensure stability in the combined heat and power network and facilitate optimal power allocation among the different energy sources.
2502.03992
Ontology-Guided, Hybrid Prompt Learning for Generalization in Knowledge Graph Question Answering
cs.CL cs.AI
Most existing Knowledge Graph Question Answering (KGQA) approaches are designed for a specific KG, such as Wikidata, DBpedia or Freebase. Due to the heterogeneity of the underlying graph schema, topology and assertions, most KGQA systems cannot be transferred to unseen Knowledge Graphs (KGs) without resource-intensive training data. We present OntoSCPrompt, a novel Large Language Model (LLM)-based KGQA approach with a two-stage architecture that separates semantic parsing from KG-dependent interactions. OntoSCPrompt first generates a SPARQL query structure (including SPARQL keywords such as SELECT, ASK, WHERE and placeholders for missing tokens) and then fills them with KG-specific information. To enhance the understanding of the underlying KG, we present an ontology-guided, hybrid prompt learning strategy that integrates KG ontology into the learning process of hybrid prompts (e.g., discrete and continuous vectors). We also present several task-specific decoding strategies to ensure the correctness and executability of generated SPARQL queries in both stages. Experimental results demonstrate that OntoSCPrompt performs as well as SOTA approaches without retraining on a number of KGQA datasets such as CWQ, WebQSP and LC-QuAD 1.0 in a resource-efficient manner and can generalize well to unseen domain-specific KGs like DBLP-QuAD and CoyPu KG Code: \href{https://github.com/LongquanJiang/OntoSCPrompt}{https://github.com/LongquanJiang/OntoSCPrompt}
2502.03997
CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing
cs.CV
Computer Aided Design (CAD) is indispensable across various industries. \emph{Text-based CAD editing}, which automates the modification of CAD models based on textual instructions, holds great potential but remains underexplored. Existing methods primarily focus on design variation generation or text-based CAD generation, either lacking support for text-based control or neglecting existing CAD models as constraints. We introduce \emph{CAD-Editor}, the first framework for text-based CAD editing. To address the challenge of demanding triplet data with accurate correspondence for training, we propose an automated data synthesis pipeline. This pipeline utilizes design variation models to generate pairs of original and edited CAD models and employs Large Vision-Language Models (LVLMs) to summarize their differences into editing instructions. To tackle the composite nature of text-based CAD editing, we propose a locate-then-infill framework that decomposes the task into two focused sub-tasks: locating regions requiring modification and infilling these regions with appropriate edits. Large Language Models (LLMs) serve as the backbone for both sub-tasks, leveraging their capabilities in natural language understanding and CAD knowledge. Experiments show that CAD-Editor achieves superior performance both quantitatively and qualitatively.
2502.03998
Online Learning of Counter Categories and Ratings in PvP Games
cs.LG cs.AI cs.GT cs.MA
In competitive games, strength ratings like Elo are widely used to quantify player skill and support matchmaking by accounting for skill disparities better than simple win rate statistics. However, scalar ratings cannot handle complex intransitive relationships, such as counter strategies seen in Rock-Paper-Scissors. To address this, recent work introduced Neural Rating Table and Neural Counter Table, which combine scalar ratings with discrete counter categories to model intransitivity. While effective, these methods rely on neural network training and cannot perform real-time updates. In this paper, we propose an online update algorithm that extends Elo principles to incorporate real-time learning of counter categories. Our method dynamically adjusts both ratings and counter relationships after each match, preserving the explainability of scalar ratings while addressing intransitivity. Experiments on zero-sum competitive games demonstrate its practicality, particularly in scenarios without complex team compositions.
2502.03999
A Self-supervised Multimodal Deep Learning Approach to Differentiate Post-radiotherapy Progression from Pseudoprogression in Glioblastoma
eess.IV cs.CV
Accurate differentiation of pseudoprogression (PsP) from True Progression (TP) following radiotherapy (RT) in glioblastoma (GBM) patients is crucial for optimal treatment planning. However, this task remains challenging due to the overlapping imaging characteristics of PsP and TP. This study therefore proposes a multimodal deep-learning approach utilizing complementary information from routine anatomical MR images, clinical parameters, and RT treatment planning information for improved predictive accuracy. The approach utilizes a self-supervised Vision Transformer (ViT) to encode multi-sequence MR brain volumes to effectively capture both global and local context from the high dimensional input. The encoder is trained in a self-supervised upstream task on unlabeled glioma MRI datasets from the open BraTS2021, UPenn-GBM, and UCSF-PDGM datasets to generate compact, clinically relevant representations from FLAIR and T1 post-contrast sequences. These encoded MR inputs are then integrated with clinical data and RT treatment planning information through guided cross-modal attention, improving progression classification accuracy. This work was developed using two datasets from different centers: the Burdenko Glioblastoma Progression Dataset (n = 59) for training and validation, and the GlioCMV progression dataset from the University Hospital Erlangen (UKER) (n = 20) for testing. The proposed method achieved an AUC of 75.3%, outperforming the current state-of-the-art data-driven approaches. Importantly, the proposed approach relies on readily available anatomical MRI sequences, clinical data, and RT treatment planning information, enhancing its clinical feasibility. The proposed approach addresses the challenge of limited data availability for PsP and TP differentiation and could allow for improved clinical decision-making and optimized treatment plans for GBM patients.
2502.04004
Near-optimal Regret Using Policy Optimization in Online MDPs with Aggregate Bandit Feedback
cs.LG
We study online finite-horizon Markov Decision Processes with adversarially changing loss and aggregate bandit feedback (a.k.a full-bandit). Under this type of feedback, the agent observes only the total loss incurred over the entire trajectory, rather than the individual losses at each intermediate step within the trajectory. We introduce the first Policy Optimization algorithms for this setting. In the known-dynamics case, we achieve the first \textit{optimal} regret bound of $\tilde \Theta(H^2\sqrt{SAK})$, where $K$ is the number of episodes, $H$ is the episode horizon, $S$ is the number of states, and $A$ is the number of actions. In the unknown dynamics case we establish regret bound of $\tilde O(H^3 S \sqrt{AK})$, significantly improving the best known result by a factor of $H^2 S^5 A^2$.
2502.04008
Automating a Complete Software Test Process Using LLMs: An Automotive Case Study
cs.SE cs.AI
Vehicle API testing verifies whether the interactions between a vehicle's internal systems and external applications meet expectations, ensuring that users can access and control various vehicle functions and data. However, this task is inherently complex, requiring the alignment and coordination of API systems, communication protocols, and even vehicle simulation systems to develop valid test cases. In practical industrial scenarios, inconsistencies, ambiguities, and interdependencies across various documents and system specifications pose significant challenges. This paper presents a system designed for the automated testing of in-vehicle APIs. By clearly defining and segmenting the testing process, we enable Large Language Models (LLMs) to focus on specific tasks, ensuring a stable and controlled testing workflow. Experiments conducted on over 100 APIs demonstrate that our system effectively automates vehicle API testing. The results also confirm that LLMs can efficiently handle mundane tasks requiring human judgment, making them suitable for complete automation in similar industrial contexts.
2502.04012
Malleable Robots
cs.RO
This chapter is about the fundamentals of fabrication, control, and human-robot interaction of a new type of collaborative robotic manipulators, called malleable robots, which are based on adjustable architectures of varying stiffness for achieving high dexterity with lower mobility arms. Collaborative robots, or cobots, commonly integrate six or more degrees of freedom (DOF) in a serial arm in order to allow positioning in constrained spaces and adaptability across tasks. Increasing the dexterity of robotic arms has been indeed traditionally accomplished by increasing the number of degrees of freedom of the system; however, once a robotic task has been established (e.g., a pick-and-place operation), the motion of the end-effector can be normally achieved using less than 6-DOF (i.e., lower mobility). The aim of malleable robots is to close the technological gap that separates current cobots from achieving flexible, accessible manufacturing automation with a reduced number of actuators.
2502.04014
Enhancing people localisation in drone imagery for better crowd management by utilising every pixel in high-resolution images
cs.CV cs.RO
Accurate people localisation using drones is crucial for effective crowd management, not only during massive events and public gatherings but also for monitoring daily urban crowd flow. Traditional methods for tiny object localisation using high-resolution drone imagery often face limitations in precision and efficiency, primarily due to constraints in image scaling and sliding window techniques. To address these challenges, a novel approach dedicated to point-oriented object localisation is proposed. Along with this approach, the Pixel Distill module is introduced to enhance the processing of high-definition images by extracting spatial information from individual pixels at once. Additionally, a new dataset named UP-COUNT, tailored to contemporary drone applications, is shared. It addresses a wide range of challenges in drone imagery, such as simultaneous camera and object movement during the image acquisition process, pushing forward the capabilities of crowd management applications. A comprehensive evaluation of the proposed method on the proposed dataset and the commonly used DroneCrowd dataset demonstrates the superiority of our approach over existing methods and highlights its efficacy in drone-based crowd object localisation tasks. These improvements markedly increase the algorithm's applicability to operate in real-world scenarios, enabling more reliable localisation and counting of individuals in dynamic environments.
2502.04018
PINT: Physics-Informed Neural Time Series Models with Applications to Long-term Inference on WeatherBench 2m-Temperature Data
cs.LG
This paper introduces PINT (Physics-Informed Neural Time Series Models), a framework that integrates physical constraints into neural time series models to improve their ability to capture complex dynamics. We apply PINT to the ERA5 WeatherBench dataset, focusing on long-term forecasting of 2m-temperature data. PINT incorporates the Simple Harmonic Oscillator Equation as a physics-informed prior, embedding its periodic dynamics into RNN, LSTM, and GRU architectures. This equation's analytical solutions (sine and cosine functions) facilitate rigorous evaluation of the benefits of incorporating physics-informed constraints. By benchmarking against a linear regression baseline derived from its exact solutions, we quantify the impact of embedding physical principles in data-driven models. Unlike traditional time series models that rely on future observations, PINT is designed for practical forecasting. Using only the first 90 days of observed data, it iteratively predicts the next two years, addressing challenges posed by limited real-time updates. Experiments on the WeatherBench dataset demonstrate PINT's ability to generalize, capture periodic trends, and align with physical principles. This study highlights the potential of physics-informed neural models in bridging machine learning and interpretable climate applications. Our models and datasets are publicly available on GitHub: https://github.com/KV-Park.
2502.04021
Variational Quantum Optimization with Continuous Bandits
cs.LG quant-ph
We introduce a novel approach to variational Quantum algorithms (VQA) via continuous bandits. VQA are a class of hybrid Quantum-classical algorithms where the parameters of Quantum circuits are optimized by classical algorithms. Previous work has used zero and first order gradient based methods, however such algorithms suffer from the barren plateau (BP) problem where gradients and loss differences are exponentially small. We introduce an approach using bandits methods which combine global exploration with local exploitation. We show how VQA can be formulated as a best arm identification problem in a continuous space of arms with Lipschitz smoothness. While regret minimization has been addressed in this setting, existing methods for pure exploration only cover discrete spaces. We give the first results for pure exploration in a continuous setting and derive a fixed-confidence, information-theoretic, instance specific lower bound. Under certain assumptions on the expected payoff, we derive a simple algorithm, which is near-optimal with respect to our lower bound. Finally, we apply our continuous bandit algorithm to two VQA schemes: a PQC and a QAOA quantum circuit, showing that we significantly outperform the previously known state of the art methods (which used gradient based methods).
2502.04022
Quantification of Biodiversity from Historical Survey Text with LLM-based Best-Worst Scaling
cs.CL
In this study, we evaluate methods to determine the frequency of species via quantity estimation from historical survey text. To that end, we formulate classification tasks and finally show that this problem can be adequately framed as a regression task using Best-Worst Scaling (BWS) with Large Language Models (LLMs). We test Ministral-8B, DeepSeek-V3, and GPT-4, finding that the latter two have reasonable agreement with humans and each other. We conclude that this approach is more cost-effective and similarly robust compared to a fine-grained multi-class approach, allowing automated quantity estimation across species.
2502.04024
A Robust Optimization Model for Cost-Efficient and Fast Electric Vehicle Charging with L2-norm Uncertainty
eess.SY cs.SY
In this paper, we propose a robust optimization model that addresses both the cost-efficiency and fast charging requirements for electric vehicles (EVs) at charging stations. By combining elements from traditional cost-minimization models and a fast charging objective, we construct an optimization model that balances user costs with rapid power allocation. Additionally, we incorporate L2-norm uncertainty into the charging cost, ensuring that the model remains resilient under cost fluctuations. The proposed model is tested under real-world scenarios and demonstrates its potential for efficient and flexible EV charging solutions.
2502.04028
Deep Meta Coordination Graphs for Multi-agent Reinforcement Learning
cs.LG
This paper presents deep meta coordination graphs (DMCG) for learning cooperative policies in multi-agent reinforcement learning (MARL). Coordination graph formulations encode local interactions and accordingly factorize the joint value function of all agents to improve efficiency in MARL. However, existing approaches rely solely on pairwise relations between agents, which potentially oversimplifies complex multi-agent interactions. DMCG goes beyond these simple direct interactions by also capturing useful higher-order and indirect relationships among agents. It generates novel graph structures accommodating multiple types of interactions and arbitrary lengths of multi-hop connections in coordination graphs to model such interactions. It then employs a graph convolutional network module to learn powerful representations in an end-to-end manner. We demonstrate its effectiveness in multiple coordination problems in MARL where other state-of-the-art methods can suffer from sample inefficiency or fail entirely. All codes can be found here: https://github.com/Nikunj-Gupta/dmcg-marl.
2502.04030
Fine, I'll Merge It Myself: A Multi-Fidelity Framework for Automated Model Merging
cs.AI cs.LG
Reasoning capabilities represent a critical frontier for large language models (LLMs), but developing them requires extensive proprietary datasets and computational resources. One way to efficiently supplement capabilities with is by model merging, which offers a promising alternative by combining multiple models without retraining. However, current merging approaches rely on manually-designed strategies for merging hyperparameters, limiting the exploration of potential model combinations and requiring significant human effort. We propose an Automated Model Merging Framework that enables fine-grained exploration of merging strategies while reducing costs through multi-fidelity approximations. We support both single and multi-objective optimization and introduce two novel search spaces: layerwise fusion (LFS) and depth-wise integration (DIS). Evaluating across a number of benchmarks, we find that the search autonomously finds 1) Merges that further boost single-objective performance, even on tasks the model has already been finetuned on, and 2) Merges that optimize multi-objective frontiers across tasks. Effective merges are found with limited compute, e.g. within less than 500 search steps.
2502.04034
Generalize Drug Response Prediction by Latent Independent Projection for Asymmetric Constrained Domain Generalization
cs.LG cs.AI
The accurate prediction of drug responses remains a formidable challenge, particularly at the single-cell level and in clinical treatment contexts. Some studies employ transfer learning techniques to predict drug responses in individual cells and patients, but they require access to target-domain data during training, which is often unavailable or only obtainable in future. In this study, we propose a novel domain generalization framework, termed panCancerDR, to address this challenge. We conceptualize each cancer type as a distinct source domain, with its cell lines serving as domain-specific samples. Our primary objective is to extract domain-invariant features from the expression profiles of cell lines across diverse cancer types, thereby generalize the predictive capacity to out-of-distribution samples. To enhance robustness, we introduce a latent independence projection (LIP) module that encourages the encoder to extract informative yet non-redundant features. Also, we propose an asymmetric adaptive clustering constraint, which clusters drug-sensitive samples into a compact group while drives resistant samples dispersed across separate clusters in the latent space. Our empirical experiments demonstrate that panCancerDR effectively learns task-relevant features from diverse source domains, and achieves accurate predictions of drug response for unseen cancer type during training. Furthermore, when evaluated on single-cell and patient-level prediction tasks, our model-trained solely on in vitro cell line data without access to target-domain information-consistently outperforms and matched current state-of-the-art methods. These findings highlights the potential of our method for real-world clinical applications.
2502.04037
Exploring Imbalanced Annotations for Effective In-Context Learning
cs.CL cs.LG
Large language models (LLMs) have shown impressive performance on downstream tasks through in-context learning (ICL), which heavily relies on the demonstrations selected from annotated datasets. Existing selection methods may hinge on the distribution of annotated datasets, which can often be long-tailed in real-world scenarios. In this work, we show that imbalanced class distributions in annotated datasets significantly degrade the performance of ICL across various tasks and selection methods. Moreover, traditional rebalance methods fail to ameliorate the issue of class imbalance in ICL. Our method is motivated by decomposing the distributional differences between annotated and test datasets into two-component weights: class-wise weights and conditional bias. The key idea behind our method is to estimate the conditional bias by minimizing the empirical error on a balanced validation dataset and to employ the two-component weights to modify the original scoring functions during selection. Our approach can prevent selecting too many demonstrations from a single class while preserving the effectiveness of the original selection methods. Extensive experiments demonstrate the effectiveness of our method, improving the average accuracy by up to 5.46 on common benchmarks with imbalanced datasets.
2502.04038
Simulating the Emergence of Differential Case Marking with Communicating Neural-Network Agents
cs.CL
Differential Case Marking (DCM) refers to the phenomenon where grammatical case marking is applied selectively based on semantic, pragmatic, or other factors. The emergence of DCM has been studied in artificial language learning experiments with human participants, which were specifically aimed at disentangling the effects of learning from those of communication (Smith & Culbertson, 2020). Multi-agent reinforcement learning frameworks based on neural networks have gained significant interest to simulate the emergence of human-like linguistic phenomena. In this study, we employ such a framework in which agents first acquire an artificial language before engaging in communicative interactions, enabling direct comparisons to human result. Using a very generic communication optimization algorithm and neural-network learners that have no prior experience with language or semantic preferences, our results demonstrate that learning alone does not lead to DCM, but when agents communicate, differential use of markers arises. This supports Smith and Culbertson (2020)'s findings that highlight the critical role of communication in shaping DCM and showcases the potential of neural-agent models to complement experimental research on language evolution.
2502.04039
A Cloud-native Agile approach to cyber platform prototyping and integration for astronomy: the ENGAGE SKA case
astro-ph.IM cs.SY eess.SY physics.med-ph
The Square Kilometre Array (SKA) Observatory is gearing up the formal construction of its two radio interferometers in Australia and South Africa after the end of design and pre-construction phases. Agile methodologies, the Cloud native Computing technologies and the DevOps software ideas are influencing the design of compute infrastructures that will be key to reduce the operational costs of SKA while improving the control and monitoring of the SKA antennas and ancillary systems, Correlators, HPC facilities or related data centre tiered systems. These tools will likely include advanced power metering technologies and efficient distribution automation and Network Operation Centres (NOC). SKA will become the world's largest radio telescope and is expected to achieve its first science by 2026. To cope with this dimension and complexity, a key part of this distributed Observatory is the overall software control and monitoring system embodied in the Observatory Management and Control (OMC) and the Services Teams that requires specialized Agile Teams to assist in software and cyber infrastructure building using an Agile development environment that includes test automation, Continuous Integration, and Continuous Deployment. To manage such a large and distributed machine, the Agile approach was adopted for the core software package of the SKA Telescope aimed at scheduling observations, controlling their execution, monitoring the telescope status and ensuring scalability and reliability. Here, we report on the ENGAGE SKA ciberinfrastructure prototyping support to the SKA Agile Software Development Life Cycle (SDLC).
2502.04040
Leveraging Reasoning with Guidelines to Elicit and Utilize Knowledge for Enhancing Safety Alignment
cs.LG cs.AI cs.CL
Training safe LLMs is one of the most critical research challenge. However, the commonly used method, Refusal Training (RT), struggles to generalize against various OOD jailbreaking attacks. Many safety training methods have been proposed to address this issue. While they offer valuable insights, we aim to complement this line of research by investigating whether OOD attacks truly exceed the capability of RT model. Conducting evaluation with BoN, we observe significant improvements on generalization as N increases. This underscores that the model possesses sufficient safety-related latent knowledge, but RT fails to consistently elicit this knowledge when addressing OOD attacks. Further analysis based on domain adaptation reveals that training with direct refusal causes model to rely on superficial shortcuts, resulting in learning of non-robust representation mappings. Based on our findings, we propose training model to perform safety reasoning for each query. Reasoning supervision encourages model to perform more computations, explicitly eliciting and using latent knowledge through reasoning. To achieve this, we synthesize reasoning supervision based on pre-guidelines, training the model to reason in alignment with them, thereby effectively eliciting and utilizing latent knowledge from diverse perspectives. Extensive experiments show that our method significantly improves generalization performance against OOD attacks.
2502.04043
Probe-Free Low-Rank Activation Intervention
cs.LG cs.AI
Language models (LMs) can produce texts that appear accurate and coherent but contain untruthful or toxic content. Inference-time interventions that edit the hidden activations have shown promising results in steering the LMs towards desirable generations. Existing activation intervention methods often comprise an activation probe to detect undesirable generation, triggering the activation modification to steer subsequent generation. This paper proposes a probe-free intervention method FLORAIN for all attention heads in a specific activation layer. It eliminates the need to train classifiers for probing purposes. The intervention function is parametrized by a sample-wise nonlinear low-rank mapping, which is trained by minimizing the distance between the modified activations and their projection onto the manifold of desirable content. Under specific constructions of the manifold and projection distance, we show that the intervention strategy can be computed efficiently by solving a smooth optimization problem. The empirical results, benchmarked on multiple base models, demonstrate that FLORAIN consistently outperforms several baseline methods in enhancing model truthfulness and quality across generation and multiple-choice tasks.
2502.04045
Comparing privacy notions for protection against reconstruction attacks in machine learning
cs.LG cs.CR cs.IT math.IT
Within the machine learning community, reconstruction attacks are a principal concern and have been identified even in federated learning (FL), which was designed with privacy preservation in mind. In response to these threats, the privacy community recommends the use of differential privacy (DP) in the stochastic gradient descent algorithm, termed DP-SGD. However, the proliferation of variants of DP in recent years\textemdash such as metric privacy\textemdash has made it challenging to conduct a fair comparison between different mechanisms due to the different meanings of the privacy parameters $\epsilon$ and $\delta$ across different variants. Thus, interpreting the practical implications of $\epsilon$ and $\delta$ in the FL context and amongst variants of DP remains ambiguous. In this paper, we lay a foundational framework for comparing mechanisms with differing notions of privacy guarantees, namely $(\epsilon,\delta)$-DP and metric privacy. We provide two foundational means of comparison: firstly, via the well-established $(\epsilon,\delta)$-DP guarantees, made possible through the R\'enyi differential privacy framework; and secondly, via Bayes' capacity, which we identify as an appropriate measure for reconstruction threats.
2502.04048
Adaptive Output Feedback MPC with Guaranteed Stability and Robustness
eess.SY cs.SY math.OC
This work proposes an adaptive output feedback model predictive control (MPC) framework for uncertain systems subject to external disturbances. In the absence of exact knowledge about the plant parameters and complete state measurements, the MPC optimization problem is reformulated in terms of their estimates derived from a suitably designed robust adaptive observer. The MPC routine returns a homothetic tube for the state estimate trajectory. Sets that characterize the state estimation errors are then added to the homothetic tube sections, resulting in a larger tube containing the true state trajectory. The two-tier tube architecture provides robustness to uncertainties due to imperfect parameter knowledge, external disturbances, and incomplete state information. Additionally, recursive feasibility and robust exponential stability are guaranteed and validated using a numerical example.
2502.04050
PartEdit: Fine-Grained Image Editing using Pre-Trained Diffusion Models
cs.CV
We present the first text-based image editing approach for object parts based on pre-trained diffusion models. Diffusion-based image editing approaches capitalized on the deep understanding of diffusion models of image semantics to perform a variety of edits. However, existing diffusion models lack sufficient understanding of many object parts, hindering fine-grained edits requested by users. To address this, we propose to expand the knowledge of pre-trained diffusion models to allow them to understand various object parts, enabling them to perform fine-grained edits. We achieve this by learning special textual tokens that correspond to different object parts through an efficient token optimization process. These tokens are optimized to produce reliable localization masks at each inference step to localize the editing region. Leveraging these masks, we design feature-blending and adaptive thresholding strategies to execute the edits seamlessly. To evaluate our approach, we establish a benchmark and an evaluation protocol for part editing. Experiments show that our approach outperforms existing editing methods on all metrics and is preferred by users 77-90% of the time in conducted user studies.
2502.04052
Decision Trees That Remember: Gradient-Based Learning of Recurrent Decision Trees with Memory
cs.LG
Neural architectures such as Recurrent Neural Networks (RNNs), Transformers, and State-Space Models have shown great success in handling sequential data by learning temporal dependencies. Decision Trees (DTs), on the other hand, remain a widely used class of models for structured tabular data but are typically not designed to capture sequential patterns directly. Instead, DT-based approaches for time-series data often rely on feature engineering, such as manually incorporating lag features, which can be suboptimal for capturing complex temporal dependencies. To address this limitation, we introduce ReMeDe Trees, a novel recurrent DT architecture that integrates an internal memory mechanism, similar to RNNs, to learn long-term dependencies in sequential data. Our model learns hard, axis-aligned decision rules for both output generation and state updates, optimizing them efficiently via gradient descent. We provide a proof-of-concept study on synthetic benchmarks to demonstrate the effectiveness of our approach.
2502.04054
Precision Agriculture Revolution: Integrating Digital Twins and Advanced Crop Recommendation for Optimal Yield
cs.LG
With the help of a digital twin structure, Agriculture 4.0 technologies like weather APIs (Application programming interface), GPS (Global Positioning System) modules, and NPK (Nitrogen, Phosphorus and Potassium) soil sensors and machine learning recommendation models, we seek to revolutionize agricultural production through this concept. In addition to providing precise crop growth forecasts, the combination of real-time data on soil composition, meteorological dynamics, and geographic coordinates aims to support crop recommendation models and simulate predictive scenarios for improved water and pesticide management.
2502.04055
Evaluating Inter-Column Logical Relationships in Synthetic Tabular Data Generation
cs.LG
Current evaluations of synthetic tabular data mainly focus on how well joint distributions are modeled, often overlooking the assessment of their effectiveness in preserving realistic event sequences and coherent entity relationships across columns.This paper proposes three evaluation metrics designed to assess the preservation of logical relationships among columns in synthetic tabular data. We validate these metrics by assessing the performance of both classical and state-of-the-art generation methods on a real-world industrial dataset.Experimental results reveal that existing methods often fail to rigorously maintain logical consistency (e.g., hierarchical relationships in geography or organization) and dependencies (e.g., temporal sequences or mathematical relationships), which are crucial for preserving the fine-grained realism of real-world tabular data. Building on these insights, this study also discusses possible pathways to better capture logical relationships while modeling the distribution of synthetic tabular data.
2502.04056
TQ-DiT: Efficient Time-Aware Quantization for Diffusion Transformers
cs.LG eess.SP
Diffusion transformers (DiTs) combine transformer architectures with diffusion models. However, their computational complexity imposes significant limitations on real-time applications and sustainability of AI systems. In this study, we aim to enhance the computational efficiency through model quantization, which represents the weights and activation values with lower precision. Multi-region quantization (MRQ) is introduced to address the asymmetric distribution of network values in DiT blocks by allocating two scaling parameters to sub-regions. Additionally, time-grouping quantization (TGQ) is proposed to reduce quantization error caused by temporal variation in activations. The experimental results show that the proposed algorithm achieves performance comparable to the original full-precision model with only a 0.29 increase in FID at W8A8. Furthermore, it outperforms other baselines at W6A6, thereby confirming its suitability for low-bit quantization. These results highlight the potential of our method to enable efficient real-time generative models.
2502.04057
Smart IoT Security: Lightweight Machine Learning Techniques for Multi-Class Attack Detection in IoT Networks
cs.LG
In the growing terrain of the Internet of Things (IoT), it is vital that networks are secure to protect against a range of cyber threats. Based on the strong machine learning framework, this study proposes novel lightweight ensemble approaches for improving multi-class attack detection of IoT devices. Using the large CICIoT 2023 dataset with 34 attack types distributed amongst 10 attack categories, we systematically evaluated the performance of a wide variety of modern machine learning methods with the aim of establishing the best-performing algorithmic choice to secure IoT applications. In particular, we explore approaches based on ML classifiers to tackle the biocharges characterized by the challenging and heterogeneous nature of attack vectors in IoT environments. The method that performed best was the Decision Tree, with an accuracy of 99.56% and an F1 score of 99.62%, showing that this model is capable of accurately and reliably detecting threats.The Random Forest model was the next best-performing model with 98.22% and an F1 score of 98.24%, suggesting that ML methods are quite effective in a situation of high-dimensional data. Our results highlight the potential for using ML classifiers in bolstering security for IoT devices and also serve as motivations for future investigations targeting scalable, keystroke-based attack detection systems. We believe that our method provides a new path to develop complex machine learning algorithms for low-resource IoT devices, balancing both accuracy and time efficiency needs. In summary, these contributions enrich the state of the art of the IoT security literature, laying down solid ground and guidelines for the deployment of smart, adaptive security in IoT settings.
2502.04058
Strategic Learning with Local Explanations as Feedback
cs.AI
We investigate algorithmic decision problems where agents can respond strategically to the decision maker's (DM) models. The demand for clear and actionable explanations from DMs to (potentially strategic) agents continues to rise. While prior work often treats explanations as full model disclosures, explanations in practice might convey only partial information, which can lead to misinterpretations and harmful responses. When full disclosure of the predictive model is neither feasible nor desirable, a key open question is how DMs can use explanations to maximise their utility without compromising agent welfare. In this work, we explore well-known local and global explanation methods, and establish a necessary condition to prevent explanations from misleading agents into self-harming actions. Moreover, with conditional homogeneity, we establish that action recommendation (AR)-based explanations are sufficient for non-harmful responses, akin to the revelation principle in information design. To operationalise AR-based explanations, we propose a simple algorithm to jointly optimise the predictive model and AR policy to balance DM outcomes with agent welfare. Our empirical results demonstrate the benefits of this approach as a more refined strategy for safe and effective partial model disclosure in algorithmic decision-making.
2502.04062
On Sufficient Richness for Linear Time-Invariant Systems
eess.SY cs.SY
Persistent excitation (PE) is a necessary and sufficient condition for uniform exponential parameter convergence in several adaptive, identification, and learning schemes. In this article, we consider, in the context of multi-input linear time-invariant (LTI) systems, the problem of guaranteeing PE of commonly-used regressors by applying a sufficiently rich (SR) input signal. Exploiting the analogies between time shifts and time derivatives, we state simple necessary and sufficient PE conditions for the discrete- and continuous-time frameworks. Moreover, we characterize the shape of the set of SR input signals for both single-input and multi-input systems. Finally, we show with a numerical example that the derived conditions are tight and cannot be improved without including additional knowledge of the considered LTI system.
2502.04064
Inteligencia artificial para la multi-clasificaci\'on de fauna en fotograf\'ias autom\'aticas utilizadas en investigaci\'on cient\'ifica
cs.CV
The management of natural environments, whether for conservation or production, requires a deep understanding of wildlife. The number, location, and behavior of wild animals are among the main subjects of study in ecology and wildlife research. The use of camera traps offers the opportunity to quickly collect large quantities of photographs that capture wildlife in its natural habitat, avoiding factors that could alter their behavior. In Tierra del Fuego, Argentina, research is being conducted on forest use by different herbivores (guanacos, cows, sheep) to optimize management and protect these natural ecosystems. Although camera traps allow for the collection of millions of images, interpreting such photographs presents a scalability challenge for manual processing. As a result, much of the valuable knowledge stored in these vast data repositories remains untapped. Neural Networks and Deep Learning are areas of study within Artificial Intelligence. Over the past decade, these two disciplines have made significant contributions to image recognition on a global scale. Ecological and wildlife conservation studies can be combined with these new technologies to extract important information from the photographs obtained by camera traps, contributing to the understanding of various natural processes and improving the management of the involved wild areas. Our project aims to develop neural network models to classify animal species in photographs taken with camera traps, addressing large-scale challenges in scientific research.
2502.04066
Predicting Large Language Model Capabilities on Closed-Book QA Tasks Using Only Information Available Prior to Training
cs.CL cs.AI
The GPT-4 technical report from OpenAI suggests that model performance on specific tasks can be predicted prior to training, though methodologies remain unspecified. This approach is crucial for optimizing resource allocation and ensuring data alignment with target tasks. To achieve this vision, we focus on predicting performance on Closed-book Question Answering (CBQA) tasks, which are closely tied to pre-training data and knowledge retention. We address three major challenges: 1) mastering the entire pre-training process, especially data construction; 2) evaluating a model's knowledge retention; and 3) predicting task-specific knowledge retention using only information available prior to training. To tackle these challenges, we pre-train three large language models (i.e., 1.6B, 7B, and 13B) using 560k dollars and 520k GPU hours. We analyze the pre-training data with knowledge triples and assess knowledge retention using established methods. Additionally, we introduce the SMI metric, an information-theoretic measure that quantifies the relationship between pre-training data, model size, and task-specific knowledge retention. Our experiments reveal a strong linear correlation ($\text{R}^2 > 0.84$) between the SMI metric and the model's accuracy on CBQA tasks across models of varying sizes (i.e., 1.1B, 1.6B, 7B, and 13B). The dataset, model, and code are available at https://github.com/yuhui1038/SMI.
2502.04074
3D Prior is All You Need: Cross-Task Few-shot 2D Gaze Estimation
cs.CV
3D and 2D gaze estimation share the fundamental objective of capturing eye movements but are traditionally treated as two distinct research domains. In this paper, we introduce a novel cross-task few-shot 2D gaze estimation approach, aiming to adapt a pre-trained 3D gaze estimation network for 2D gaze prediction on unseen devices using only a few training images. This task is highly challenging due to the domain gap between 3D and 2D gaze, unknown screen poses, and limited training data. To address these challenges, we propose a novel framework that bridges the gap between 3D and 2D gaze. Our framework contains a physics-based differentiable projection module with learnable parameters to model screen poses and project 3D gaze into 2D gaze. The framework is fully differentiable and can integrate into existing 3D gaze networks without modifying their original architecture. Additionally, we introduce a dynamic pseudo-labelling strategy for flipped images, which is particularly challenging for 2D labels due to unknown screen poses. To overcome this, we reverse the projection process by converting 2D labels to 3D space, where flipping is performed. Notably, this 3D space is not aligned with the camera coordinate system, so we learn a dynamic transformation matrix to compensate for this misalignment. We evaluate our method on MPIIGaze, EVE, and GazeCapture datasets, collected respectively on laptops, desktop computers, and mobile devices. The superior performance highlights the effectiveness of our approach, and demonstrates its strong potential for real-world applications.
2502.04075
Controllable Emotion Generation with Emotion Vectors
cs.CL
In recent years, technologies based on large-scale language models (LLMs) have made remarkable progress in many fields, especially in customer service, content creation, and embodied intelligence, showing broad application potential. However, The LLM's ability to express emotions with proper tone, timing, and in both direct and indirect forms is still insufficient but significant. Few works have studied on how to build the controlable emotional expression capability of LLMs. In this work, we propose a method for emotion expression output by LLMs, which is universal, highly flexible, and well controllable proved with the extensive experiments and verifications. This method has broad application prospects in fields involving emotions output by LLMs, such as intelligent customer service, literary creation, and home companion robots. The extensive experiments on various LLMs with different model-scales and architectures prove the versatility and the effectiveness of the proposed method.
2502.04076
Content-Rich AIGC Video Quality Assessment via Intricate Text Alignment and Motion-Aware Consistency
cs.CV
The advent of next-generation video generation models like \textit{Sora} poses challenges for AI-generated content (AIGC) video quality assessment (VQA). These models substantially mitigate flickering artifacts prevalent in prior models, enable longer and complex text prompts and generate longer videos with intricate, diverse motion patterns. Conventional VQA methods designed for simple text and basic motion patterns struggle to evaluate these content-rich videos. To this end, we propose \textbf{CRAVE} (\underline{C}ontent-\underline{R}ich \underline{A}IGC \underline{V}ideo \underline{E}valuator), specifically for the evaluation of Sora-era AIGC videos. CRAVE proposes the multi-granularity text-temporal fusion that aligns long-form complex textual semantics with video dynamics. Additionally, CRAVE leverages the hybrid motion-fidelity modeling to assess temporal artifacts. Furthermore, given the straightforward prompts and content in current AIGC VQA datasets, we introduce \textbf{CRAVE-DB}, a benchmark featuring content-rich videos from next-generation models paired with elaborate prompts. Extensive experiments have shown that the proposed CRAVE achieves excellent results on multiple AIGC VQA benchmarks, demonstrating a high degree of alignment with human perception. All data and code will be publicly available at https://github.com/littlespray/CRAVE.
2502.04077
AttentionPredictor: Temporal Pattern Matters for Efficient LLM Inference
cs.CL cs.LG
With the development of large language models (LLMs), efficient inference through Key-Value (KV) cache compression has attracted considerable attention, especially for long-context generation. To compress the KV cache, recent methods identify critical KV tokens through heuristic ranking with attention scores. However, these methods often struggle to accurately determine critical tokens as they neglect the \textit{temporal patterns} in attention scores, resulting in a noticeable degradation in LLM performance. To address this challenge, we propose AttentionPredictor, which is the first learning-based critical token identification approach. Specifically, AttentionPredictor learns a lightweight convolution model to capture spatiotemporal patterns and predict the next-token attention score. An appealing feature of AttentionPredictor is that it accurately predicts the attention score while consuming negligible memory. Moreover, we propose a cross-token critical cache prefetching framework that hides the token estimation time overhead to accelerate the decoding stage. By retaining most of the attention information, AttentionPredictor achieves 16$\times$ KV cache compression with comparable LLM performance, significantly outperforming the state-of-the-art.
2502.04079
DEALing with Image Reconstruction: Deep Attentive Least Squares
eess.IV cs.CV cs.LG
State-of-the-art image reconstruction often relies on complex, highly parameterized deep architectures. We propose an alternative: a data-driven reconstruction method inspired by the classic Tikhonov regularization. Our approach iteratively refines intermediate reconstructions by solving a sequence of quadratic problems. These updates have two key components: (i) learned filters to extract salient image features, and (ii) an attention mechanism that locally adjusts the penalty of filter responses. Our method achieves performance on par with leading plug-and-play and learned regularizer approaches while offering interpretability, robustness, and convergent behavior. In effect, we bridge traditional regularization and deep learning with a principled reconstruction approach.
2502.04083
Automatic quantification of breast cancer biomarkers from multiple 18F-FDG PET image segmentation
cs.CV cs.AI
Neoadjuvant chemotherapy (NAC) has become a standard clinical practice for tumor downsizing in breast cancer with 18F-FDG Positron Emission Tomography (PET). Our work aims to leverage PET imaging for the segmentation of breast lesions. The focus is on developing an automated system that accurately segments primary tumor regions and extracts key biomarkers from these areas to provide insights into the evolution of breast cancer following the first course of NAC. 243 baseline 18F-FDG PET scans (PET_Bl) and 180 follow-up 18F-FDG PET scans (PET_Fu) were acquired before and after the first course of NAC, respectively. Firstly, a deep learning-based breast tumor segmentation method was developed. The optimal baseline model (model trained on baseline exams) was fine-tuned on 15 follow-up exams and adapted using active learning to segment tumor areas in PET_Fu. The pipeline computes biomarkers such as maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) to evaluate tumor evolution between PET_Fu and PET_Bl. Quality control measures were employed to exclude aberrant outliers. The nnUNet deep learning model outperformed in tumor segmentation on PET_Bl, achieved a Dice similarity coefficient (DSC) of 0.89 and a Hausdorff distance (HD) of 3.52 mm. After fine-tuning, the model demonstrated a DSC of 0.78 and a HD of 4.95 mm on PET_Fu exams. Biomarkers analysis revealed very strong correlations whatever the biomarker between manually segmented and automatically predicted regions. The significant average decrease of SUVmax, MTV and TLG were 5.22, 11.79 cm3 and 19.23 cm3, respectively. The presented approach demonstrates an automated system for breast tumor segmentation from 18F-FDG PET. Thanks to the extracted biomarkers, our method enables the automatic assessment of cancer progression.
2502.04088
Quantifying imperfect cognition via achieved information gain
cs.IT math.IT
Cognition, the process of information processing in form of inference, communication, and memorization, is the central activity of any intelligence. Its physical realization in a brain, computer, or in any other intelligent system requires resources like time, energy, memory, bandwidth, money, and others. Due to limited resources, many real world intelligent systems perform only imperfect cognition. For understanding the trade-off between accuracy and resource investments in existing systems, e.g. in biology, as well as for the resource-aware optimal design of information processing systems, like computer algorithms and artificial neural networks, a quantification of information obtained in an imperfect cognitive operation is desirable. To this end, we propose the concept of achieved information gain (AIG) of a belief update, which is given by the amount of information obtained by updating from the initial knowledge state to the ideal one, minus the amount a change from the imperfect to the ideal state would yield. AIG has many properties desired for quantifying imperfect cognition. The ratio of achieved to ideally obtainable information measures cognitive fidelity and that of AIG to the necessary cognitive effort measures cognitive efficiency. We provide an axiomatic derivation of AIG, illustrate its application at common scenarios of posterior inaccuracies, and discuss the implication of cognitive efficiency for sustainable resource allocation in computational inference.
2502.04094
Soft and Highly-Integrated Optical Fiber Bending Sensors for Proprioception in Multi-Material 3D Printed Fingers
cs.RO
Accurate shape sensing, only achievable through distributed proprioception, is a key requirement for closed-loop control of soft robots. Low-cost power efficient optoelectronic sensors manufactured from flexible materials represent a natural choice as they can cope with the large deformations of soft robots without loss of performance. However, existing integration approaches are cumbersome and require manual steps and complex assembly. We propose a semi-automated printing process where plastic optical fibers are embedded with readout electronics in 3D printed flexures. The fibers become locked in place and the readout electronics remain optically coupled to them while the flexures undergo large bending deformations, creating a repeatable, monolithically manufactured bending transducer with only 10 minutes required in total for the manual embedding steps. We demonstrate the process by manufacturing multi-material 3D printed fingers and extensively evaluating the performance of each proprioceptive joint. The sensors achieve 70% linearity and 4.81{\deg} RMS error on average. Furthermore, the distributed architecture allows for maintaining an average fingertip position estimation accuracy of 12 mm in the presence of external static forces. To demonstrate the potential of the distributed sensor architecture in robotics applications, we build a data-driven model independent of actuation feedback to detect contact with objects in the environment.
2502.04095
LLMs to Support a Domain Specific Knowledge Assistant
cs.CL cs.AI
This work presents a custom approach to developing a domain specific knowledge assistant for sustainability reporting using the International Financial Reporting Standards (IFRS). In this domain, there is no publicly available question-answer dataset, which has impeded the development of a high-quality chatbot to support companies with IFRS reporting. The two key contributions of this project therefore are: (1) A high-quality synthetic question-answer (QA) dataset based on IFRS sustainability standards, created using a novel generation and evaluation pipeline leveraging Large Language Models (LLMs). This comprises 1,063 diverse QA pairs that address a wide spectrum of potential user queries in sustainability reporting. Various LLM-based techniques are employed to create the dataset, including chain-of-thought reasoning and few-shot prompting. A custom evaluation framework is developed to assess question and answer quality across multiple dimensions, including faithfulness, relevance, and domain specificity. The dataset averages a score range of 8.16 out of 10 on these metrics. (2) Two architectures for question-answering in the sustainability reporting domain - a RAG pipeline and a fully LLM-based pipeline. The architectures are developed by experimenting, fine-tuning, and training on the QA dataset. The final pipelines feature an LLM fine-tuned on domain specific data and an industry classification component to improve the handling of complex queries. The RAG architecture achieves an accuracy of 85.32% on single-industry and 72.15% on cross-industry multiple-choice questions, outperforming the baseline approach by 4.67 and 19.21 percentage points, respectively. The LLM-based pipeline achieves an accuracy of 93.45% on single-industry and 80.30% on cross-industry multiple-choice questions, an improvement of 12.80 and 27.36 percentage points over the baseline, respectively.
2502.04098
Efficient Few-Shot Continual Learning in Vision-Language Models
cs.CV cs.AI
Vision-language models (VLMs) excel in tasks such as visual question answering and image captioning. However, VLMs are often limited by their use of pretrained image encoders, like CLIP, leading to image understanding errors that hinder overall performance. On top of that, real-world applications often require the model to be continuously adapted as new and often limited data continuously arrive. To address this, we propose LoRSU (Low-Rank Adaptation with Structured Updates), a robust and computationally efficient method for selectively updating image encoders within VLMs. LoRSU introduces structured and localized parameter updates, effectively correcting performance on previously error-prone data while preserving the model's general robustness. Our approach leverages theoretical insights to identify and update only the most critical parameters, achieving significant resource efficiency. Specifically, we demonstrate that LoRSU reduces computational overhead by over 25x compared to full VLM updates, without sacrificing performance. Experimental results on VQA tasks in the few-shot continual learning setting, validate LoRSU's scalability, efficiency, and effectiveness, making it a compelling solution for image encoder adaptation in resource-constrained environments.
2502.04101
Safe Quadrotor Navigation using Composite Control Barrier Functions
cs.RO
This paper introduces a safety filter to ensure collision avoidance for multirotor aerial robots. The proposed formalism leverages a single Composite Control Barrier Function from all position constraints acting on a third-order nonlinear representation of the robot's dynamics. We analyze the recursive feasibility of the safety filter under the composite constraint and demonstrate that the infeasible set is negligible. The proposed method allows computational scalability against thousands of constraints and, thus, complex scenes with numerous obstacles. We experimentally demonstrate its ability to guarantee the safety of a quadrotor with an onboard LiDAR, operating in both indoor and outdoor cluttered environments against both naive and adversarial nominal policies.
2502.04103
VTutor: An Open-Source SDK for Generative AI-Powered Animated Pedagogical Agents with Multi-Media Output
cs.HC cs.AI cs.SE
The rapid evolution of large language models (LLMs) has transformed human-computer interaction (HCI), but the interaction with LLMs is currently mainly focused on text-based interactions, while other multi-model approaches remain under-explored. This paper introduces VTutor, an open-source Software Development Kit (SDK) that combines generative AI with advanced animation technologies to create engaging, adaptable, and realistic APAs for human-AI multi-media interactions. VTutor leverages LLMs for real-time personalized feedback, advanced lip synchronization for natural speech alignment, and WebGL rendering for seamless web integration. Supporting various 2D and 3D character models, VTutor enables researchers and developers to design emotionally resonant, contextually adaptive learning agents. This toolkit enhances learner engagement, feedback receptivity, and human-AI interaction while promoting trustworthy AI principles in education. VTutor sets a new standard for next-generation APAs, offering an accessible, scalable solution for fostering meaningful and immersive human-AI interaction experiences. The VTutor project is open-sourced and welcomes community-driven contributions and showcases.
2502.04110
Ancient Greek Technology: An Immersive Learning Use Case Described Using a Co-Intelligent Custom ChatGPT Assistant
cs.HC cs.AI
Achieving consistency in immersive learning case descriptions is essential but challenging due to variations in research focus, methodology, and researchers' background. We address these challenges by leveraging the Immersive Learning Case Sheet (ILCS), a methodological instrument to standardize case descriptions, that we applied to an immersive learning case on ancient Greek technology in VRChat. Research team members had differing levels of familiarity with the ILCS and the case content, so we developed a custom ChatGPT assistant to facilitate consistent terminology and process alignment across the team. This paper constitutes an example of how structured case reports can be a novel contribution to immersive learning literature. Our findings demonstrate how the ILCS supports structured reflection and interpretation of the case. Further we report that the use of a ChatGPT assistant significantly sup-ports the coherence and quality of the team members development of the final ILCS. This exposes the potential of employing AI-driven tools to enhance collaboration and standardization of research practices in qualitative educational research. However, we also discuss the limitations and challenges, including reliance on AI for interpretive tasks and managing varied levels of expertise within the team. This study thus provides insights into the practical application of AI in standardizing immersive learning research processes.
2502.04111
Adaptive Margin Contrastive Learning for Ambiguity-aware 3D Semantic Segmentation
cs.CV
In this paper, we propose an adaptive margin contrastive learning method for 3D point cloud semantic segmentation, namely AMContrast3D. Most existing methods use equally penalized objectives, which ignore per-point ambiguities and less discriminated features stemming from transition regions. However, as highly ambiguous points may be indistinguishable even for humans, their manually annotated labels are less reliable, and hard constraints over these points would lead to sub-optimal models. To address this, we design adaptive objectives for individual points based on their ambiguity levels, aiming to ensure the correctness of low-ambiguity points while allowing mistakes for high-ambiguity points. Specifically, we first estimate ambiguities based on position embeddings. Then, we develop a margin generator to shift decision boundaries for contrastive feature embeddings, so margins are narrowed due to increasing ambiguities with even negative margins for extremely high-ambiguity points. Experimental results on large-scale datasets, S3DIS and ScanNet, demonstrate that our method outperforms state-of-the-art methods.
2502.04115
A Neural Network-based Multi-timestep Command Governor for Nonlinear Systems with Constraints
eess.SY cs.SY
The multi-timestep command governor (MCG) is an add-on algorithm that enforces constraints by modifying, at each timestep, the reference command to a pre-stabilized control system. The MCG can be interpreted as a Model-Predictive Control scheme operating on the reference command. The implementation of MCG on nonlinear systems carries a heavy computational burden as it requires solving a nonlinear program with multiple decision variables at each timestep. This paper proposes a less computationally demanding alternative, based on approximating the MCG control law using a neural network (NN) trained on offline data. However, since the NN output may not always be constraint-admissible due to training errors, its output is adjusted using a sensitivity-based method. We thus refer to the resulting control strategy as the neural network-based MCG (NN-MCG). As validation, the proposed controller is applied as a load governor for constraint management in an automotive fuel cell system. It is shown that the proposed strategy is significantly more computationally efficient than the traditional MCG, while achieving nearly identical performance if the NN is well-trained.
2502.04116
Generative Adversarial Networks Bridging Art and Machine Intelligence
cs.LG cs.CV
Generative Adversarial Networks (GAN) have greatly influenced the development of computer vision and artificial intelligence in the past decade and also connected art and machine intelligence together. This book begins with a detailed introduction to the fundamental principles and historical development of GANs, contrasting them with traditional generative models and elucidating the core adversarial mechanisms through illustrative Python examples. The text systematically addresses the mathematical and theoretical underpinnings including probability theory, statistics, and game theory providing a solid framework for understanding the objectives, loss functions, and optimisation challenges inherent to GAN training. Subsequent chapters review classic variants such as Conditional GANs, DCGANs, InfoGAN, and LAPGAN before progressing to advanced training methodologies like Wasserstein GANs, GANs with gradient penalty, least squares GANs, and spectral normalisation techniques. The book further examines architectural enhancements and task-specific adaptations in generators and discriminators, showcasing practical implementations in high resolution image generation, artistic style transfer, video synthesis, text to image generation and other multimedia applications. The concluding sections offer insights into emerging research trends, including self-attention mechanisms, transformer-based generative models, and a comparative analysis with diffusion models, thus charting promising directions for future developments in both academic and applied settings.
2502.04121
Optimizing Perturbations for Improved Training of Machine Learning Models
cs.LG cond-mat.dis-nn physics.chem-ph
Machine learning models have become indispensable tools in applications across the physical sciences. Their training is often time-consuming, vastly exceeding the inference timescales. Several protocols have been developed to perturb the learning process and improve the training, such as shrink and perturb, warm restarts, and stochastic resetting. For classifiers, these perturbations have been shown to result in enhanced speedups or improved generalization. However, the design of such perturbations is usually done \textit{ad hoc} by intuition and trial and error. To rationally optimize training protocols, we frame them as first-passage processes and consider their response to perturbations. We show that if the unperturbed learning process reaches a quasi-steady state, the response at a single perturbation frequency can predict the behavior at a wide range of frequencies. We demonstrate that this is the case when training a CIFAR-10 classifier using the ResNet-18 model and use this approach to identify an optimal perturbation and frequency. Our work allows optimization of training protocols of machine learning models using a statistical mechanical approach.
2502.04126
RC Measurement Uncertainty Estimation Method for Directive Antennas and Turntable Stirring
eess.SY cs.SY
This paper investigates measurement uncertainty in a Reverberation Chamber (RC) within the lower FR2 bands (24.25-29.5 GHz). The study focuses on the impact of several factors contributing to RC measurement uncertainty, including finite sample size, polarization imbalance, and spatial non-uniformity. A series of 24 measurements were conducted using a horn antenna, known for its directivity in mmWave frequencies, varying antenna parameters such as height, orientation, position on the turntable, and polarization within a predefined chamber volume. The measurement uncertainty was evaluated by a method based on the standardized 3GPP and CTIA approaches, incorporating uncorrelated measurements and analyzing Pearson correlation coefficients between measurement pairs. An analysis of variance (ANOVA) was performed on the frequency-averaged power transfer function to identify the significance and impact of each variable on measurement variability. Additionally, the K-factor was estimated for each measurement set as part of the RC characterization, using an alternative approach to account for the turntable stirring effect. The findings highlight which variables most significantly influence measurement uncertainty, where the antenna orientation emerges as the most significant factor for the mmWave directive antenna setup.
2502.04128
Llasa: Scaling Train-Time and Inference-Time Compute for Llama-based Speech Synthesis
eess.AS cs.AI cs.CL cs.MM cs.SD
Recent advances in text-based large language models (LLMs), particularly in the GPT series and the o1 model, have demonstrated the effectiveness of scaling both training-time and inference-time compute. However, current state-of-the-art TTS systems leveraging LLMs are often multi-stage, requiring separate models (e.g., diffusion models after LLM), complicating the decision of whether to scale a particular model during training or testing. This work makes the following contributions: First, we explore the scaling of train-time and inference-time compute for speech synthesis. Second, we propose a simple framework Llasa for speech synthesis that employs a single-layer vector quantizer (VQ) codec and a single Transformer architecture to fully align with standard LLMs such as Llama. Our experiments reveal that scaling train-time compute for Llasa consistently improves the naturalness of synthesized speech and enables the generation of more complex and accurate prosody patterns. Furthermore, from the perspective of scaling inference-time compute, we employ speech understanding models as verifiers during the search, finding that scaling inference-time compute shifts the sampling modes toward the preferences of specific verifiers, thereby improving emotional expressiveness, timbre consistency, and content accuracy. In addition, we released the checkpoint and training code for our TTS model (1B, 3B, 8B) and codec model publicly available.
2502.04131
On the importance of structural identifiability for machine learning with partially observed dynamical systems
cs.LG
The successful application of modern machine learning for time series classification is often hampered by limitations in quality and quantity of available training data. To overcome these limitations, available domain expert knowledge in the form of parametrised mechanistic dynamical models can be used whenever it is available and time series observations may be represented as an element from a given class of parametrised dynamical models. This makes the learning process interpretable and allows the modeller to deal with sparsely and irregularly sampled data in a natural way. However, the internal processes of a dynamical model are often only partially observed. This can lead to ambiguity regarding which particular model realization best explains a given time series observation. This problem is well-known in the literature, and a dynamical model with this issue is referred to as structurally unidentifiable. Training a classifier that incorporates knowledge about a structurally unidentifiable dynamical model can negatively influence classification performance. To address this issue, we employ structural identifiability analysis to explicitly relate parameter configurations that are associated with identical system outputs. Using the derived relations in classifier training, we demonstrate that this method significantly improves the classifier's ability to generalize to unseen data on a number of example models from the biomedical domain. This effect is especially pronounced when the number of training instances is limited. Our results demonstrate the importance of accounting for structural identifiability, a topic that has received relatively little attention from the machine learning community.
2502.04132
Transfer Learning for Covert Speech Classification Using EEG Hilbert Envelope and Temporal Fine Structure
cs.LG
Brain-Computer Interfaces (BCIs) can decode imagined speech from neural activity. However, these systems typically require extensive training sessions where participants imaginedly repeat words, leading to mental fatigue and difficulties identifying the onset of words, especially when imagining sequences of words. This paper addresses these challenges by transferring a classifier trained in overt speech data to covert speech classification. We used electroencephalogram (EEG) features derived from the Hilbert envelope and temporal fine structure, and used them to train a bidirectional long-short-term memory (BiLSTM) model for classification. Our method reduces the burden of extensive training and achieves state-of-the-art classification accuracy: 86.44% for overt speech and 79.82% for covert speech using the overt speech classifier.
2502.04134
The Order Effect: Investigating Prompt Sensitivity in Closed-Source LLMs
cs.CL
As large language models (LLMs) become integral to diverse applications, ensuring their reliability under varying input conditions is crucial. One key issue affecting this reliability is order sensitivity, wherein slight variations in input arrangement can lead to inconsistent or biased outputs. Although recent advances have reduced this sensitivity, the problem remains unresolved. This paper investigates the extent of order sensitivity in closed-source LLMs by conducting experiments across multiple tasks, including paraphrasing, relevance judgment, and multiple-choice questions. Our results show that input order significantly affects performance across tasks, with shuffled inputs leading to measurable declines in output accuracy. Few-shot prompting demonstrates mixed effectiveness and offers partial mitigation, however, fails to fully resolve the problem. These findings highlight persistent risks, particularly in high-stakes applications, and point to the need for more robust LLMs or improved input-handling techniques in future development.
2502.04139
Beyond the Final Layer: Hierarchical Query Fusion Transformer with Agent-Interpolation Initialization for 3D Instance Segmentation
cs.CV
3D instance segmentation aims to predict a set of object instances in a scene and represent them as binary foreground masks with corresponding semantic labels. Currently, transformer-based methods are gaining increasing attention due to their elegant pipelines, reduced manual selection of geometric properties, and superior performance. However, transformer-based methods fail to simultaneously maintain strong position and content information during query initialization. Additionally, due to supervision at each decoder layer, there exists a phenomenon of object disappearance with the deepening of layers. To overcome these hurdles, we introduce Beyond the Final Layer: Hierarchical Query Fusion Transformer with Agent-Interpolation Initialization for 3D Instance Segmentation (BFL). Specifically, an Agent-Interpolation Initialization Module is designed to generate resilient queries capable of achieving a balance between foreground coverage and content learning. Additionally, a Hierarchical Query Fusion Decoder is designed to retain low overlap queries, mitigating the decrease in recall with the deepening of layers. Extensive experiments on ScanNetV2, ScanNet200, ScanNet++ and S3DIS datasets demonstrate the superior performance of BFL.
2502.04140
Synthetic Datasets for Machine Learning on Spatio-Temporal Graphs using PDEs
cs.LG cs.AI
Many physical processes can be expressed through partial differential equations (PDEs). Real-world measurements of such processes are often collected at irregularly distributed points in space, which can be effectively represented as graphs; however, there are currently only a few existing datasets. Our work aims to make advancements in the field of PDE-modeling accessible to the temporal graph machine learning community, while addressing the data scarcity problem, by creating and utilizing datasets based on PDEs. In this work, we create and use synthetic datasets based on PDEs to support spatio-temporal graph modeling in machine learning for different applications. More precisely, we showcase three equations to model different types of disasters and hazards in the fields of epidemiology, atmospheric particles, and tsunami waves. Further, we show how such created datasets can be used by benchmarking several machine learning models on the epidemiological dataset. Additionally, we show how pre-training on this dataset can improve model performance on real-world epidemiological data. The presented methods enable others to create datasets and benchmarks customized to individual requirements. The source code for our methodology and the three created datasets can be found on https://github.com/github-usr-ano/Temporal_Graph_Data_PDEs.
2502.04141
Behavioral Entropy-Guided Dataset Generation for Offline Reinforcement Learning
cs.LG
Entropy-based objectives are widely used to perform state space exploration in reinforcement learning (RL) and dataset generation for offline RL. Behavioral entropy (BE), a rigorous generalization of classical entropies that incorporates cognitive and perceptual biases of agents, was recently proposed for discrete settings and shown to be a promising metric for robotic exploration problems. In this work, we propose using BE as a principled exploration objective for systematically generating datasets that provide diverse state space coverage in complex, continuous, potentially high-dimensional domains. To achieve this, we extend the notion of BE to continuous settings, derive tractable $k$-nearest neighbor estimators, provide theoretical guarantees for these estimators, and develop practical reward functions that can be used with standard RL methods to learn BE-maximizing policies. Using standard MuJoCo environments, we experimentally compare the performance of offline RL algorithms for a variety of downstream tasks on datasets generated using BE, R\'{e}nyi, and Shannon entropy-maximizing policies, as well as the SMM and RND algorithms. We find that offline RL algorithms trained on datasets collected using BE outperform those trained on datasets collected using Shannon entropy, SMM, and RND on all tasks considered, and on 80% of the tasks compared to datasets collected using R\'{e}nyi entropy.