id
stringlengths
9
16
title
stringlengths
4
278
categories
stringlengths
5
104
abstract
stringlengths
6
4.09k
2501.18170
Continually Evolved Multimodal Foundation Models for Cancer Prognosis
cs.LG
Cancer prognosis is a critical task that involves predicting patient outcomes and survival rates. To enhance prediction accuracy, previous studies have integrated diverse data modalities, such as clinical notes, medical images, and genomic data, leveraging their complementary information. However, existing approaches face two major limitations. First, they struggle to incorporate newly arrived data with varying distributions into training, such as patient records from different hospitals, thus rendering sub-optimal generalizability and limited utility in real-world applications. Second, most multimodal integration methods rely on simplistic concatenation or task-specific pipelines, which fail to capture the complex interdependencies across modalities. To address these, we propose a continually evolving multi-modal foundation model. Extensive experiments on the TCGA dataset demonstrate the effectiveness of our approach, highlighting its potential to advance cancer prognosis by enabling robust and adaptive multimodal integration.
2501.18174
Advancing Personalized Federated Learning: Integrative Approaches with AI for Enhanced Privacy and Customization
cs.LG eess.SP
In the age of data-driven decision making, preserving privacy while providing personalized experiences has become paramount. Personalized Federated Learning (PFL) offers a promising framework by decentralizing the learning process, thus ensuring data privacy and reducing reliance on centralized data repositories. However, the integration of advanced Artificial Intelligence (AI) techniques within PFL remains underexplored. This paper proposes a novel approach that enhances PFL with cutting-edge AI methodologies including adaptive optimization, transfer learning, and differential privacy. We present a model that not only boosts the performance of individual client models but also ensures robust privacy-preserving mechanisms and efficient resource utilization across heterogeneous networks. Empirical results demonstrate significant improvements in model accuracy and personalization, along with stringent privacy adherence, as compared to conventional federated learning models. This work paves the way for a new era of truly personalized and privacy-conscious AI systems, offering significant implications for industries requiring compliance with stringent data protection regulations.
2501.18177
Investigating Tax Evasion Emergence Using Dual Large Language Model and Deep Reinforcement Learning Powered Agent-based Simulation
cs.IR cs.CY cs.MA
Tax evasion, usually the largest component of an informal economy, is a persistent challenge over history with significant socio-economic implications. Many socio-economic studies investigate its dynamics, including influencing factors, the role and influence of taxation policies, and the prediction of the tax evasion volume over time. These studies assumed such behavior is given, as observed in the real world, neglecting the "big bang" of such activity in a population. To this end, computational economy studies adopted developments in computer simulations, in general, and recent innovations in artificial intelligence (AI), in particular, to simulate and study informal economy appearance in various socio-economic settings. This study presents a novel computational framework to examine the dynamics of tax evasion and the emergence of informal economic activity. Employing an agent-based simulation powered by Large Language Models and Deep Reinforcement Learning, the framework is uniquely designed to allow informal economic behaviors to emerge organically, without presupposing their existence or explicitly signaling agents about the possibility of evasion. This provides a rigorous approach for exploring the socio-economic determinants of compliance behavior. The experimental design, comprising model validation and exploratory phases, demonstrates the framework's robustness in replicating theoretical economic behaviors. Findings indicate that individual personality traits, external narratives, enforcement probabilities, and the perceived efficiency of public goods provision significantly influence both the timing and extent of informal economic activity. The results underscore that efficient public goods provision and robust enforcement mechanisms are complementary; neither alone is sufficient to curtail informal activity effectively.
2501.18178
Estimating Multi-chirp Parameters using Curvature-guided Langevin Monte Carlo
eess.SP cs.LG stat.ML
This paper considers the problem of estimating chirp parameters from a noisy mixture of chirps. While a rich body of work exists in this area, challenges remain when extending these techniques to chirps of higher order polynomials. We formulate this as a non-convex optimization problem and propose a modified Langevin Monte Carlo (LMC) sampler that exploits the average curvature of the objective function to reliably find the minimizer. Results show that our Curvature-guided LMC (CG-LMC) algorithm is robust and succeeds even in low SNR regimes, making it viable for practical applications.
2501.18179
Tunable Multilayer Surface Plasmon Resonance Biosensor for Trace-Level Toxin Detection
eess.SY cs.SY
This paper presents a comprehensive study on a novel multilayer surface plasmon resonance (SPR) biosensor designed for detecting trace-level toxins in liquid samples with exceptional precision and efficiency. Leveraging the Kretschmann configuration, the proposed design integrates advanced two-dimensional materials, including black phosphorus (BP) and transition metal dichalcogenides (TMDs), to significantly enhance the performance metrics of the sensor. Key innovations include the optimization of sensitivity through precise material layering, minimization of full-width at half-maximum (FWHM) to improve signal resolution, and maximization of the figure of merit (FoM) for superior detection accuracy. Numerical simulations are employed to validate the structural and functional enhancements of the biosensor. The results demonstrate improved interaction between the evanescent field and the analyte, enabling detection at trace concentrations with higher specificity. This biosensor is poised to contribute to advancements in biochemical sensing, environmental monitoring, and other critical applications requiring high-sensitivity toxin detection.
2501.18183
Decentralized Projection-free Online Upper-Linearizable Optimization with Applications to DR-Submodular Optimization
math.OC cs.CC cs.LG stat.ML
We introduce a novel framework for decentralized projection-free optimization, extending projection-free methods to a broader class of upper-linearizable functions. Our approach leverages decentralized optimization techniques with the flexibility of upper-linearizable function frameworks, effectively generalizing traditional DR-submodular function optimization. We obtain the regret of $O(T^{1-\theta/2})$ with communication complexity of $O(T^{\theta})$ and number of linear optimization oracle calls of $O(T^{2\theta})$ for decentralized upper-linearizable function optimization, for any $0\le \theta \le 1$. This approach allows for the first results for monotone up-concave optimization with general convex constraints and non-monotone up-concave optimization with general convex constraints. Further, the above results for first order feedback are extended to zeroth order, semi-bandit, and bandit feedback.
2501.18184
Genetic Algorithm with Border Trades (GAB)
cs.LG cs.NE stat.CO
This paper introduces a novel approach to improving Genetic Algorithms (GA) in large or complex problem spaces by incorporating new chromosome patterns in the breeding process through border trade activities. These strategies increase chromosome diversity, preventing premature convergence and enhancing the GA's ability to explore the solution space more effectively. Empirical evidence demonstrates significant improvements in convergence behavior. This approach offers a promising pathway to addressing challenges in optimizing large or complex problem domains.
2501.18187
In-Context Learning of Polynomial Kernel Regression in Transformers with GLU Layers
cs.LG cs.AI
Transformer-based models have demonstrated remarkable ability in in-context learning (ICL), where they can adapt to unseen tasks from a prompt with a few examples, without requiring parameter updates. Recent research has provided insight into how linear Transformers can perform ICL by implementing gradient descent estimators. In particular, it has been shown that the optimal linear self-attention (LSA) mechanism can implement one step of gradient descent with respect to a linear least-squares objective when trained on random linear regression tasks. However, the theoretical understanding of ICL for nonlinear function classes remains limited. In this work, we address this gap by first showing that LSA is inherently restricted to solving linear least-squares objectives and thus, the solutions in prior works cannot readily extend to nonlinear ICL tasks. To overcome this limitation, drawing inspiration from modern architectures, we study a mechanism that combines LSA with GLU-like feed-forward layers and show that this allows the model to perform one step of gradient descent on a polynomial kernel regression. Further, we characterize the scaling behavior of the resulting Transformer model, highlighting the necessary model size to effectively handle quadratic ICL tasks. Our findings highlight the distinct roles of attention and feed-forward layers in nonlinear ICL and identify key challenges when extending ICL to nonlinear function classes.
2501.18189
Neural Network Modeling of Microstructure Complexity Using Digital Libraries
cs.LG cond-mat.mtrl-sci cs.CE nlin.PS physics.comp-ph
Microstructure evolution in matter is often modeled numerically using field or level-set solvers, mirroring the dual representation of spatiotemporal complexity in terms of pixel or voxel data, and geometrical forms in vector graphics. Motivated by this analog, as well as the structural and event-driven nature of artificial and spiking neural networks, respectively, we evaluate their performance in learning and predicting fatigue crack growth and Turing pattern development. Predictions are made based on digital libraries constructed from computer simulations, which can be replaced by experimental data to lift the mathematical overconstraints of physics. Our assessment suggests that the leaky integrate-and-fire neuron model offers superior predictive accuracy with fewer parameters and less memory usage, alleviating the accuracy-cost tradeoff in contrast to the common practices in computer vision tasks. Examination of network architectures shows that these benefits arise from its reduced weight range and sparser connections. The study highlights the capability of event-driven models in tackling problems with evolutionary bulk-phase and interface behaviors using the digital library approach.
2501.18190
Economic Rationality under Specialization: Evidence of Decision Bias in AI Agents
cs.AI
In the study by Chen et al. (2023) [01], the large language model GPT demonstrated economic rationality comparable to or exceeding the average human level in tasks such as budget allocation and risk preference. Building on this finding, this paper further incorporates specialized agents, such as biotechnology experts and economists, for a horizontal comparison to explore whether specialization can enhance or maintain economic rationality equivalent to that of GPT in similar decision-making scenarios. The results indicate that when agents invest more effort in specialized fields, their decision-making behavior is more prone to 'rationality shift,' specifically manifested as increased violations of GARP (Generalized Axiom of Revealed Preference), decreased CCEI (Critical Cost Efficiency Index), and more significant decision deviations under high-risk conditions. In contrast, GPT and more generalized basic agents maintain a more stable and consistent level of rationality across multiple tasks. This study reveals the inherent conflict between specialization and economic rationality, providing new insights for constructing AI decision-making systems that balance specialization and generalization across various scenarios.
2501.18192
Machine Learning Fairness for Depression Detection using EEG Data
cs.CV cs.LG eess.SP
This paper presents the very first attempt to evaluate machine learning fairness for depression detection using electroencephalogram (EEG) data. We conduct experiments using different deep learning architectures such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and Gated Recurrent Unit (GRU) networks across three EEG datasets: Mumtaz, MODMA and Rest. We employ five different bias mitigation strategies at the pre-, in- and post-processing stages and evaluate their effectiveness. Our experimental results show that bias exists in existing EEG datasets and algorithms for depression detection, and different bias mitigation methods address bias at different levels across different fairness measures.
2501.18196
GDformer: Going Beyond Subsequence Isolation for Multivariate Time Series Anomaly Detection
cs.LG
Unsupervised anomaly detection of multivariate time series is a challenging task, given the requirements of deriving a compact detection criterion without accessing the anomaly points. The existing methods are mainly based on reconstruction error or association divergence, which are both confined to isolated subsequences with limited horizons, hardly promising unified series-level criterion. In this paper, we propose the Global Dictionary-enhanced Transformer (GDformer) with a renovated dictionary-based cross attention mechanism to cultivate the global representations shared by all normal points in the entire series. Accordingly, the cross-attention maps reflect the correlation weights between the point and global representations, which naturally leads to the representation-wise similarity-based detection criterion. To foster more compact detection boundary, prototypes are introduced to capture the distribution of normal point-global correlation weights. GDformer consistently achieves state-of-the-art unsupervised anomaly detection performance on five real-world benchmark datasets. Further experiments validate the global dictionary has great transferability among various datasets. The code is available at https://github.com/yuppielqx/GDformer.
2501.18197
Fundamental Challenges in Evaluating Text2SQL Solutions and Detecting Their Limitations
cs.LG cs.DB
In this work, we dive into the fundamental challenges of evaluating Text2SQL solutions and highlight potential failure causes and the potential risks of relying on aggregate metrics in existing benchmarks. We identify two largely unaddressed limitations in current open benchmarks: (1) data quality issues in the evaluation data, mainly attributed to the lack of capturing the probabilistic nature of translating a natural language description into a structured query (e.g., NL ambiguity), and (2) the bias introduced by using different match functions as approximations for SQL equivalence. To put both limitations into context, we propose a unified taxonomy of all Text2SQL limitations that can lead to both prediction and evaluation errors. We then motivate the taxonomy by providing a survey of Text2SQL limitations using state-of-the-art Text2SQL solutions and benchmarks. We describe the causes of limitations with real-world examples and propose potential mitigation solutions for each category in the taxonomy. We conclude by highlighting the open challenges encountered when deploying such mitigation strategies or attempting to automatically apply the taxonomy.
2501.18199
HKAN: Hierarchical Kolmogorov-Arnold Network without Backpropagation
cs.LG cs.AI
This paper introduces the Hierarchical Kolmogorov-Arnold Network (HKAN), a novel network architecture that offers a competitive alternative to the recently proposed Kolmogorov-Arnold Network (KAN). Unlike KAN, which relies on backpropagation, HKAN adopts a randomized learning approach, where the parameters of its basis functions are fixed, and linear aggregations are optimized using least-squares regression. HKAN utilizes a hierarchical multi-stacking framework, with each layer refining the predictions from the previous one by solving a series of linear regression problems. This non-iterative training method simplifies computation and eliminates sensitivity to local minima in the loss function. Empirical results show that HKAN delivers comparable, if not superior, accuracy and stability relative to KAN across various regression tasks, while also providing insights into variable importance. The proposed approach seamlessly integrates theoretical insights with practical applications, presenting a robust and efficient alternative for neural network modeling.
2501.18200
Characterization of Permanent Magnet Synchronous Machines based on semi-analytic model reduction for drive cycle analysis
cs.CE
The characterization of an interior permanent magnet synchronous machine (IPMSM) requires numerical analysis of the nonlinear magnetic motor model in different load conditions. To obtain the case-specific best machine behavior, a strategy for the determination of stator input current amplitude and angle is employed for all possible load torques given a limited terminal current amplitude and DC bus voltage. Various losses are calculated using state of the art loss models. The electromagnetic performance of the electric machine is stored in lookup tables. These can then be used for the drive cycle analysis of the electric drive train in the design and optimization stages. To avoid the use of a dedicated mesh generator in the numerical analysis, volumetric spline-based models are suggested.With this approach, the mesh can be generated directly from the Computer Aided Design (CAD) geometry. This enables an automatic adaption of the grid following a geometry perturbation. With this the approximated solution is kept consistent over the different iterations of an overlying optimization, improving its convergence behavior.
2501.18201
Neural Operator based Reinforcement Learning for Control of first-order PDEs with Spatially-Varying State Delay
cs.AI cs.SY eess.SY
Control of distributed parameter systems affected by delays is a challenging task, particularly when the delays depend on spatial variables. The idea of integrating analytical control theory with learning-based control within a unified control scheme is becoming increasingly promising and advantageous. In this paper, we address the problem of controlling an unstable first-order hyperbolic PDE with spatially-varying delays by combining PDE backstepping control strategies and deep reinforcement learning (RL). To eliminate the assumption on the delay function required for the backstepping design, we propose a soft actor-critic (SAC) architecture incorporating a DeepONet to approximate the backstepping controller. The DeepONet extracts features from the backstepping controller and feeds them into the policy network. In simulations, our algorithm outperforms the baseline SAC without prior backstepping knowledge and the analytical controller.
2501.18202
On Scaling Neurosymbolic Programming through Guided Logical Inference
cs.AI
Probabilistic neurosymbolic learning seeks to integrate neural networks with symbolic programming. Many state-of-the-art systems rely on a reduction to the Probabilistic Weighted Model Counting Problem (PWMC), which requires computing a Boolean formula called the logical provenance.However, PWMC is \\#P-hard, and the number of clauses in the logical provenance formula can grow exponentially, creating a major bottleneck that significantly limits the applicability of PNL solutions in practice.We propose a new approach centered around an exact algorithm DPNL, that enables bypassing the computation of the logical provenance.The DPNL approach relies on the principles of an oracle and a recursive DPLL-like decomposition in order to guide and speed up logical inference.Furthermore, we show that this approach can be adapted for approximate reasoning with $\epsilon$ or $(\epsilon, \delta)$ guarantees, called ApproxDPNL.Experiments show significant performance gains.DPNL enables scaling exact inference further, resulting in more accurate models.Further, ApproxDPNL shows potential for advancing the scalability of neurosymbolic programming by incorporating approximations even further, while simultaneously ensuring guarantees for the reasoning process.
2501.18203
Joint Design and Pricing of Extended Warranties for Multiple Automobiles with Different Price Bands
math.OC cs.SY eess.SY
Extended warranties (EWs) are significant source of revenue for capital-intensive products like automobiles. Such products consist of multiple subsystems, providing flexibility in EW customization, for example, bundling a tailored set of subsystems in an EW contract. This, in turn, enables the creation of a service menu with different EW contract options. From the perspective of a third-party EW provider servicing a fleet of automobile brands, we develop a novel model to jointly optimize the design and pricing of EWs in order to maximize the profit. Specifically, the problem is to determine which contracts should be included in the EW menu and identify the appropriate price for each contract. As the complexity of the joint optimization problem increases exponentially with the number of subsystems, two solution approaches are devised to solve the problem. The first approach is based on a mixed-integer second-order cone programming reformulation, which guarantees optimality but is applicable only for a small number of subsystems. The second approach utilizes a two-step iteration process, offering enhanced computational efficiency in scenarios with a large number of subsystems. Through numerical experiments, the effectiveness of our model is validated, particularly in scenarios characterized by high failure rates and a large number of subsystems.
2501.18205
Contextually Structured Token Dependency Encoding for Large Language Models
cs.CL
Token representation strategies within large-scale neural architectures often rely on contextually refined embeddings, yet conventional approaches seldom encode structured relationships explicitly within token interactions. Self-attention mechanisms effectively capture dynamic contextual dependencies, but their reliance on learned weight distributions limits the preservation of long-range hierarchical structures in generated sequences. Dependency-aware token encoding introduces a structured approach to embedding initialization, ensuring that relational constraints are embedded within token representations rather than inferred solely through attention dynamics. The proposed encoding mechanism refines token interactions through dependency-weighted attention computations, ensuring that syntactic and semantic dependencies are retained across multiple processing layers. Empirical evaluations indicate reductions in perplexity across diverse linguistic benchmarks, suggesting improvements in contextual coherence and predictive consistency in autoregressive text generation. Computational efficiency assessments reveal a moderate increase in memory consumption and training time, attributed to additional matrix computations within the encoding module, yet scalability remains feasible within conventional transformer architectures. Structured encoding enhances lexical variation and dependency retention, reinforcing linguistic coherence without requiring external syntactic annotations or auxiliary training objectives. Statistical comparisons highlight improvements in dependency alignment, particularly in longer sequences where conventional self-attention models exhibit degradation in hierarchical consistency. Sentence length distributions indicate a reduction in abrupt phrase transitions, further supporting the hypothesis that explicit dependency encoding facilitates more structured phrase generation.
2501.18210
Hashtag Re-Appropriation for Audience Control on Recommendation-Driven Social Media Xiaohongshu (rednote)
cs.HC cs.CY cs.IR cs.SI
Algorithms have played a central role in personalized recommendations on social media. However, they also present significant obstacles for content creators trying to predict and manage their audience reach. This issue is particularly challenging for marginalized groups seeking to maintain safe spaces. Our study explores how women on Xiaohongshu (rednote), a recommendation-driven social platform, proactively re-appropriate hashtags (e.g., #Baby Supplemental Food) by using them in posts unrelated to their literal meaning. The hashtags were strategically chosen from topics that would be uninteresting to the male audience they wanted to block. Through a mixed-methods approach, we analyzed the practice of hashtag re-appropriation based on 5,800 collected posts and interviewed 24 active users from diverse backgrounds to uncover users' motivations and reactions towards the re-appropriation. This practice highlights how users can reclaim agency over content distribution on recommendation-driven platforms, offering insights into self-governance within algorithmic-centered power structures.
2501.18216
Behavior Modeling Space Reconstruction for E-Commerce Search
cs.IR
Delivering superior search services is crucial for enhancing customer experience and driving revenue growth. Conventionally, search systems model user behaviors by combining user preference and query item relevance statically, often through a fixed logical 'and' relationship. This paper reexamines existing approaches through a unified lens using both causal graphs and Venn diagrams, uncovering two prevalent yet significant issues: entangled preference and relevance effects, and a collapsed modeling space. To surmount these challenges, our research introduces a novel framework, DRP, which enhances search accuracy through two components to reconstruct the behavior modeling space. Specifically, we implement preference editing to proactively remove the relevance effect from preference predictions, yielding untainted user preferences. Additionally, we employ adaptive fusion, which dynamically adjusts fusion criteria to align with the varying patterns of relevance and preference, facilitating more nuanced and tailored behavior predictions within the reconstructed modeling space. Empirical validation on two public datasets and a proprietary search dataset underscores the superiority of our proposed methodology, demonstrating marked improvements in performance over existing approaches.
2501.18219
Revisiting $\Psi$DONet: microlocally inspired filters for incomplete-data tomographic reconstructions
math.OC cs.CV cs.LG
In this paper, we revisit a supervised learning approach based on unrolling, known as $\Psi$DONet, by providing a deeper microlocal interpretation for its theoretical analysis, and extending its study to the case of sparse-angle tomography. Furthermore, we refine the implementation of the original $\Psi$DONet considering special filters whose structure is specifically inspired by the streak artifact singularities characterizing tomographic reconstructions from incomplete data. This allows to considerably lower the number of (learnable) parameters while preserving (or even slightly improving) the same quality for the reconstructions from limited-angle data and providing a proof-of-concept for the case of sparse-angle tomographic data.
2501.18220
On-Line Learning for Planning and Control of Underactuated Robots with Uncertain Dynamics
cs.RO
We present an iterative approach for planning and controlling motions of underactuated robots with uncertain dynamics. At its core, there is a learning process which estimates the perturbations induced by the model uncertainty on the active and passive degrees of freedom. The generic iteration of the algorithm makes use of the learned data in both the planning phase, which is based on optimization, and the control phase, where partial feedback linearization of the active dofs is performed on the model updated on-line. The performance of the proposed approach is shown by comparative simulations and experiments on a Pendubot executing various types of swing-up maneuvers. Very few iterations are typically needed to generate dynamically feasible trajectories and the tracking control that guarantees their accurate execution, even in the presence of large model uncertainties.
2501.18223
Exploring Large Protein Language Models in Constrained Evaluation Scenarios within the FLIP Benchmark
cs.LG cs.AI
In this study, we expand upon the FLIP benchmark-designed for evaluating protein fitness prediction models in small, specialized prediction tasks-by assessing the performance of state-of-the-art large protein language models, including ESM-2 and SaProt on the FLIP dataset. Unlike larger, more diverse benchmarks such as ProteinGym, which cover a broad spectrum of tasks, FLIP focuses on constrained settings where data availability is limited. This makes it an ideal framework to evaluate model performance in scenarios with scarce task-specific data. We investigate whether recent advances in protein language models lead to significant improvements in such settings. Our findings provide valuable insights into the performance of large-scale models in specialized protein prediction tasks.
2501.18229
GPD: Guided Polynomial Diffusion for Motion Planning
cs.RO
Diffusion-based motion planners are becoming popular due to their well-established performance improvements, stemming from sample diversity and the ease of incorporating new constraints directly during inference. However, a primary limitation of the diffusion process is the requirement for a substantial number of denoising steps, especially when the denoising process is coupled with gradient-based guidance. In this paper, we introduce, diffusion in the parametric space of trajectories, where the parameters are represented as Bernstein coefficients. We show that this representation greatly improves the effectiveness of the cost function guidance and the inference speed. We also introduce a novel stitching algorithm that leverages the diversity in diffusion-generated trajectories to produce collision-free trajectories with just a single cost function-guided model. We demonstrate that our approaches outperform current SOTA diffusion-based motion planners for manipulators and provide an ablation study on key components.
2501.18232
Free-T2M: Frequency Enhanced Text-to-Motion Diffusion Model With Consistency Loss
cs.CV
Rapid progress in text-to-motion generation has been largely driven by diffusion models. However, existing methods focus solely on temporal modeling, thereby overlooking frequency-domain analysis. We identify two key phases in motion denoising: the **semantic planning stage** and the **fine-grained improving stage**. To address these phases effectively, we propose **Fre**quency **e**nhanced **t**ext-**to**-**m**otion diffusion model (**Free-T2M**), incorporating stage-specific consistency losses that enhance the robustness of static features and improve fine-grained accuracy. Extensive experiments demonstrate the effectiveness of our method. Specifically, on StableMoFusion, our method reduces the FID from **0.189** to **0.051**, establishing a new SOTA performance within the diffusion architecture. These findings highlight the importance of incorporating frequency-domain insights into text-to-motion generation for more precise and robust results.
2501.18236
RIS-assisted Physical Layer Security
cs.IT eess.SP math.IT
We propose a reconfigurable intelligent surface (RIS)-assisted wiretap channel, where the RIS is strategically deployed to provide a spatial separation to the transmitter, and orthogonal combiners are employed at the legitimate receiver to extract the data streams from the direct and RIS-assisted links. Then we derive the achievable secrecy rate under semantic security for the RIS-assisted channel and design an algorithm for the secrecy rate optimization problem. The simulation results show the effects of total transmit power, the location and number of eavesdroppers on the security performance.
2501.18237
Arbitrary Data as Images: Fusion of Patient Data Across Modalities and Irregular Intervals with Vision Transformers
cs.CV cs.AI
A patient undergoes multiple examinations in each hospital stay, where each provides different facets of the health status. These assessments include temporal data with varying sampling rates, discrete single-point measurements, therapeutic interventions such as medication administration, and images. While physicians are able to process and integrate diverse modalities intuitively, neural networks need specific modeling for each modality complicating the training procedure. We demonstrate that this complexity can be significantly reduced by visualizing all information as images along with unstructured text and subsequently training a conventional vision-text transformer. Our approach, Vision Transformer for irregular sampled Multi-modal Measurements (ViTiMM), not only simplifies data preprocessing and modeling but also outperforms current state-of-the-art methods in predicting in-hospital mortality and phenotyping, as evaluated on 6,175 patients from the MIMIC-IV dataset. The modalities include patient's clinical measurements, medications, X-ray images, and electrocardiography scans. We hope our work inspires advancements in multi-modal medical AI by reducing the training complexity to (visual) prompt engineering, thus lowering entry barriers and enabling no-code solutions for training. The source code will be made publicly available.
2501.18243
Statistical multi-metric evaluation and visualization of LLM system predictive performance
stat.AP cs.CL cs.LG
The evaluation of generative or discriminative large language model (LLM)-based systems is often a complex multi-dimensional problem. Typically, a set of system configuration alternatives are evaluated on one or more benchmark datasets, each with one or more evaluation metrics, which may differ between datasets. We often want to evaluate -- with a statistical measure of significance -- whether systems perform differently either on a given dataset according to a single metric, on aggregate across metrics on a dataset, or across datasets. Such evaluations can be done to support decision-making, such as deciding whether a particular system component change (e.g., choice of LLM or hyperparameter values) significantly improves performance over the current system configuration, or, more generally, whether a fixed set of system configurations (e.g., a leaderboard list) have significantly different performances according to metrics of interest. We present a framework implementation that automatically performs the correct statistical tests, properly aggregates the statistical results across metrics and datasets (a nontrivial task), and can visualize the results. The framework is demonstrated on the multi-lingual code generation benchmark CrossCodeEval, for several state-of-the-art LLMs.
2501.18246
Ground Awareness in Deep Learning for Large Outdoor Point Cloud Segmentation
cs.CV
This paper presents an analysis of utilizing elevation data to aid outdoor point cloud semantic segmentation through existing machine-learning networks in remote sensing, specifically in urban, built-up areas. In dense outdoor point clouds, the receptive field of a machine learning model may be too small to accurately determine the surroundings and context of a point. By computing Digital Terrain Models (DTMs) from the point clouds, we extract the relative elevation feature, which is the vertical distance from the terrain to a point. RandLA-Net is employed for efficient semantic segmentation of large-scale point clouds. We assess its performance across three diverse outdoor datasets captured with varying sensor technologies and sensor locations. Integration of relative elevation data leads to consistent performance improvements across all three datasets, most notably in the Hessigheim dataset, with an increase of 3.7 percentage points in average F1 score from 72.35% to 76.01%, by establishing long-range dependencies between ground and objects. We also explore additional local features such as planarity, normal vectors, and 2D features, but their efficacy varied based on the characteristics of the point cloud. Ultimately, this study underscores the important role of the non-local relative elevation feature for semantic segmentation of point clouds in remote sensing applications.
2501.18250
Dynamic Model Fine-Tuning For Extreme MIMO CSI Compression
cs.IT eess.SP math.IT
Efficient channel state information (CSI) compression is crucial in frequency division duplexing (FDD) massive multiple-input multiple-output (MIMO) systems due to excessive feedback overhead. Recently, deep learning-based compression techniques have demonstrated superior performance across various data types, including CSI. However, these approaches often experience performance degradation when the data distribution changes due to their limited generalization capabilities. To address this challenge, we propose a model fine-tuning approach for CSI feedback in massive MIMO systems. The idea is to fine-tune the encoder/decoder network models in a dynamic fashion using the recent CSI samples. First, we explore encoder-only fine-tuning, where only the encoder parameters are updated, leaving the decoder and latent parameters unchanged. Next, we consider full-model fine-tuning, where the encoder and decoder models are jointly updated. Unlike encoder-only fine-tuning, full-model fine-tuning requires the updated decoder and latent parameters to be transmitted to the decoder side. To efficiently handle this, we propose different prior distributions for model updates, such as uniform and truncated Gaussian to entropy code them together with the compressed CSI and account for additional feedback overhead imposed by conveying the model updates. Moreover, we incorporate quantized model updates during fine-tuning to reflect the impact of quantization in the deployment phase. Our results demonstrate that full-model fine-tuning significantly enhances the rate-distortion (RD) performance of neural CSI compression. Furthermore, we analyze how often the full-model fine-tuning should be applied in a new wireless environment and identify an optimal period interval for achieving the best RD trade-off.
2501.18251
How to Select Datapoints for Efficient Human Evaluation of NLG Models?
cs.CL
Human evaluation is the gold-standard for evaluating text generation models. It is also expensive, and to fit budgetary constraints, a random subset of the test data is often chosen in practice. The randomly selected data may not accurately represent test performance, making this approach economically inefficient for model comparison. Thus, in this work, we develop a suite of selectors to get the most informative datapoints for human evaluation while taking the evaluation costs into account. We show that selectors based on variance in automated metric scores, diversity in model outputs, or Item Response Theory outperform random selection. We further develop an approach to distill these selectors to the scenario where the model outputs are not yet available. In particular, we introduce source-based estimators, which predict item usefulness for human evaluation just based on the source texts. We demonstrate the efficacy of our selectors in two common NLG tasks, machine translation and summarization, and show that up to only ~50% of the test data is needed to produce the same evaluation result as the entire data. Our implementations are published in the subset2evaluate package.
2501.18258
PDE-DKL: PDE-constrained deep kernel learning in high dimensionality
cs.LG cs.AI stat.ML
Many physics-informed machine learning methods for PDE-based problems rely on Gaussian processes (GPs) or neural networks (NNs). However, both face limitations when data are scarce and the dimensionality is high. Although GPs are known for their robust uncertainty quantification in low-dimensional settings, their computational complexity becomes prohibitive as the dimensionality increases. In contrast, while conventional NNs can accommodate high-dimensional input, they often require extensive training data and do not offer uncertainty quantification. To address these challenges, we propose a PDE-constrained Deep Kernel Learning (PDE-DKL) framework that combines DL and GPs under explicit PDE constraints. Specifically, NNs learn a low-dimensional latent representation of the high-dimensional PDE problem, reducing the complexity of the problem. GPs then perform kernel regression subject to the governing PDEs, ensuring accurate solutions and principled uncertainty quantification, even when available data are limited. This synergy unifies the strengths of both NNs and GPs, yielding high accuracy, robust uncertainty estimates, and computational efficiency for high-dimensional PDEs. Numerical experiments demonstrate that PDE-DKL achieves high accuracy with reduced data requirements. They highlight its potential as a practical, reliable, and scalable solver for complex PDE-based applications in science and engineering.
2501.18265
Collecting Cost-Effective, High-Quality Truthfulness Assessments with LLM Summarized Evidence
cs.IR cs.CL cs.HC
With the degradation of guardrails against mis- and disinformation online, it is more critical than ever to be able to effectively combat it. In this paper, we explore the efficiency and effectiveness of using crowd-sourced truthfulness assessments based on condensed, large language model (LLM) generated summaries of online sources. We compare the use of generated summaries to the use of original web pages in an A/B testing setting, where we employ a large and diverse pool of crowd-workers to perform the truthfulness assessment. We evaluate the quality of assessments, the efficiency with which assessments are performed, and the behavior and engagement of participants. Our results demonstrate that the Summary modality, which relies on summarized evidence, offers no significant change in assessment accuracy over the Standard modality, while significantly increasing the speed with which assessments are performed. Workers using summarized evidence produce a significantly higher number of assessments in the same time frame, reducing the cost needed to acquire truthfulness assessments. Additionally, the Summary modality maximizes both the inter-annotator agreements as well as the reliance on and perceived usefulness of evidence, demonstrating the utility of summarized evidence without sacrificing the quality of assessments.
2501.18268
Reducing Aleatoric and Epistemic Uncertainty through Multi-modal Data Acquisition
cs.LG
To generate accurate and reliable predictions, modern AI systems need to combine data from multiple modalities, such as text, images, audio, spreadsheets, and time series. Multi-modal data introduces new opportunities and challenges for disentangling uncertainty: it is commonly assumed in the machine learning community that epistemic uncertainty can be reduced by collecting more data, while aleatoric uncertainty is irreducible. However, this assumption is challenged in modern AI systems when information is obtained from different modalities. This paper introduces an innovative data acquisition framework where uncertainty disentanglement leads to actionable decisions, allowing sampling in two directions: sample size and data modality. The main hypothesis is that aleatoric uncertainty decreases as the number of modalities increases, while epistemic uncertainty decreases by collecting more observations. We provide proof-of-concept implementations on two multi-modal datasets to showcase our data acquisition framework, which combines ideas from active learning, active feature acquisition and uncertainty quantification.
2501.18269
MAMS: Model-Agnostic Module Selection Framework for Video Captioning
cs.CV cs.AI
Multi-modal transformers are rapidly gaining attention in video captioning tasks. Existing multi-modal video captioning methods typically extract a fixed number of frames, which raises critical challenges. When a limited number of frames are extracted, important frames with essential information for caption generation may be missed. Conversely, extracting an excessive number of frames includes consecutive frames, potentially causing redundancy in visual tokens extracted from consecutive video frames. To extract an appropriate number of frames for each video, this paper proposes the first model-agnostic module selection framework in video captioning that has two main functions: (1) selecting a caption generation module with an appropriate size based on visual tokens extracted from video frames, and (2) constructing subsets of visual tokens for the selected caption generation module. Furthermore, we propose a new adaptive attention masking scheme that enhances attention on important visual tokens. Our experiments on three different benchmark datasets demonstrate that the proposed framework significantly improves the performance of three recent video captioning models.
2501.18270
The iToBoS dataset: skin region images extracted from 3D total body photographs for lesion detection
eess.IV cs.AI cs.CV
Artificial intelligence has significantly advanced skin cancer diagnosis by enabling rapid and accurate detection of malignant lesions. In this domain, most publicly available image datasets consist of single, isolated skin lesions positioned at the center of the image. While these lesion-centric datasets have been fundamental for developing diagnostic algorithms, they lack the context of the surrounding skin, which is critical for improving lesion detection. The iToBoS dataset was created to address this challenge. It includes 16,954 images of skin regions from 100 participants, captured using 3D total body photography. Each image roughly corresponds to a $7 \times 9$ cm section of skin with all suspicious lesions annotated using bounding boxes. Additionally, the dataset provides metadata such as anatomical location, age group, and sun damage score for each image. This dataset aims to facilitate training and benchmarking of algorithms, with the goal of enabling early detection of skin cancer and deployment of this technology in non-clinical environments.
2501.18271
Pre-Trained Vision-Language Model Selection and Reuse for Downstream Tasks
cs.LG cs.AI
Pre-trained Vision-Language Models (VLMs) are becoming increasingly popular across various visual tasks, and several open-sourced VLM variants have been released. However, selecting the best-performing pre-trained VLM for a specific downstream task is challenging since no single VLM can achieve promising performance on all downstream tasks, and evaluating all available VLMs is impossible due to time and data limitations. To address this problem, this paper proposes a novel paradigm to select and reuse VLM for downstream tasks, called Model Label Learning (MLL). The proposal contains three key modules: \emph{model labeling}, which assigns labels to each VLM to describe their specialty and utility; \emph{model selection}, which matches the requirements of the target task with model labels; and \emph{model reuse}, which applies selected VLMs to the target task in an ensemble manner. The proposal is highly computationally efficient and growable since the model labeling process is completed target task independent and the ability could grow with the number of candidate VLMs. We also introduce a new benchmark for evaluating VLM selection methods, including 49 VLMs and 17 target task datasets. Experimental results clearly demonstrate the effectiveness of the proposed method for selecting and reusing VLMs.
2501.18277
Sebra: Debiasing Through Self-Guided Bias Ranking
cs.LG
Ranking samples by fine-grained estimates of spuriosity (the degree to which spurious cues are present) has recently been shown to significantly benefit bias mitigation, over the traditional binary biased-\textit{vs}-unbiased partitioning of train sets. However, this spuriosity ranking comes with the requirement of human supervision. In this paper, we propose a debiasing framework based on our novel \ul{Se}lf-Guided \ul{B}ias \ul{Ra}nking (\emph{Sebra}), that mitigates biases (spurious correlations) via an automatic ranking of data points by spuriosity within their respective classes. Sebra leverages a key local symmetry in Empirical Risk Minimization (ERM) training -- the ease of learning a sample via ERM inversely correlates with its spuriousity; the fewer spurious correlations a sample exhibits, the harder it is to learn, and vice versa. However, globally across iterations, ERM tends to deviate from this symmetry. Sebra dynamically steers ERM to correct this deviation, facilitating the sequential learning of attributes in increasing order of difficulty, \ie, decreasing order of spuriosity. As a result, the sequence in which Sebra learns samples naturally provides spuriousity rankings. We use the resulting fine-grained bias characterization in a contrastive learning framework to mitigate biases from multiple sources. Extensive experiments show that Sebra consistently outperforms previous state-of-the-art unsupervised debiasing techniques across multiple standard benchmarks, including UrbanCars, BAR, CelebA, and ImageNet-1K. Code, pre-trained models, and training logs are available at https://kadarsh22.github.io/sebra_iclr25/.
2501.18278
ReactEmbed: A Cross-Domain Framework for Protein-Molecule Representation Learning via Biochemical Reaction Networks
cs.LG
The challenge in computational biology and drug discovery lies in creating comprehensive representations of proteins and molecules that capture their intrinsic properties and interactions. Traditional methods often focus on unimodal data, such as protein sequences or molecular structures, limiting their ability to capture complex biochemical relationships. This work enhances these representations by integrating biochemical reactions encompassing interactions between molecules and proteins. By leveraging reaction data alongside pre-trained embeddings from state-of-the-art protein and molecule models, we develop ReactEmbed, a novel method that creates a unified embedding space through contrastive learning. We evaluate ReactEmbed across diverse tasks, including drug-target interaction, protein-protein interaction, protein property prediction, and molecular property prediction, consistently surpassing all current state-of-the-art models. Notably, we showcase ReactEmbed's practical utility through successful implementation in lipid nanoparticle-based drug delivery, enabling zero-shot prediction of blood-brain barrier permeability for protein-nanoparticle complexes. The code and comprehensive database of reaction pairs are available for open use at \href{https://github.com/amitaysicherman/ReactEmbed}{GitHub}.
2501.18280
Jailbreaking LLMs' Safeguard with Universal Magic Words for Text Embedding Models
cs.CL cs.AI cs.LG cs.NE
The security issue of large language models (LLMs) has gained significant attention recently, with various defense mechanisms developed to prevent harmful outputs, among which safeguards based on text embedding models serve as a fundamental defense. Through testing, we discover that the distribution of text embedding model outputs is significantly biased with a large mean. Inspired by this observation, we propose novel efficient methods to search for universal magic words that can attack text embedding models. The universal magic words as suffixes can move the embedding of any text towards the bias direction, therefore manipulate the similarity of any text pair and mislead safeguards. By appending magic words to user prompts and requiring LLMs to end answers with magic words, attackers can jailbreak the safeguard. To eradicate this security risk, we also propose defense mechanisms against such attacks, which can correct the biased distribution of text embeddings in a train-free manner.
2501.18282
Leveraging Sparsity for Sample-Efficient Preference Learning: A Theoretical Perspective
cs.LG
This paper considers the sample-efficiency of preference learning, which models and predicts human choices based on comparative judgments. The minimax optimal estimation rate $\Theta(d/n)$ in traditional estimation theory requires that the number of samples $n$ scales linearly with the dimensionality of the feature space $d$. However, the high dimensionality of the feature space and the high cost of collecting human-annotated data challenge the efficiency of traditional estimation methods. To remedy this, we leverage sparsity in the preference model and establish sharp estimation rates. We show that under the sparse random utility model, where the parameter of the reward function is $k$-sparse, the minimax optimal rate can be reduced to $\Theta(k/n \log(d/k))$. Furthermore, we analyze the $\ell_{1}$-regularized estimator and show that it achieves near-optimal rate under mild assumptions on the Gram matrix. Experiments on synthetic data and LLM alignment data validate our theoretical findings, showing that sparsity-aware methods significantly reduce sample complexity and improve prediction accuracy.
2501.18283
Random Feature Representation Boosting
stat.ML cs.LG
We introduce Random Feature Representation Boosting (RFRBoost), a novel method for constructing deep residual random feature neural networks (RFNNs) using boosting theory. RFRBoost uses random features at each layer to learn the functional gradient of the network representation, enhancing performance while preserving the convex optimization benefits of RFNNs. In the case of MSE loss, we obtain closed-form solutions to greedy layer-wise boosting with random features. For general loss functions, we show that fitting random feature residual blocks reduces to solving a quadratically constrained least squares problem. We demonstrate, through numerical experiments on 91 tabular datasets for regression and classification, that RFRBoost significantly outperforms traditional RFNNs and end-to-end trained MLP ResNets, while offering substantial computational advantages and theoretical guarantees stemming from boosting theory.
2501.18287
Mining for Species, Locations, Habitats, and Ecosystems from Scientific Papers in Invasion Biology: A Large-Scale Exploratory Study with Large Language Models
cs.CL cs.AI cs.DL
This paper presents an exploratory study that harnesses the capabilities of large language models (LLMs) to mine key ecological entities from invasion biology literature. Specifically, we focus on extracting species names, their locations, associated habitats, and ecosystems, information that is critical for understanding species spread, predicting future invasions, and informing conservation efforts. Traditional text mining approaches often struggle with the complexity of ecological terminology and the subtle linguistic patterns found in these texts. By applying general-purpose LLMs without domain-specific fine-tuning, we uncover both the promise and limitations of using these models for ecological entity extraction. In doing so, this study lays the groundwork for more advanced, automated knowledge extraction tools that can aid researchers and practitioners in understanding and managing biological invasions.
2501.18288
Normalizing flows for SU($N$) gauge theories employing singular value decomposition
hep-lat cs.LG
We present a progress report on the use of normalizing flows for generating gauge field configurations in pure SU(N) gauge theories. We discuss how the singular value decomposition can be used to construct gauge-invariant quantities, which serve as the building blocks for designing gauge-equivariant transformations of SU(N) gauge links. Using this novel approach, we build representative models for the SU(3) Wilson action on a \( 4^4 \) lattice with \( \beta = 1 \). We train these models and provide an analysis of their performance, highlighting the effectiveness of the new technique for gauge-invariant transformations. We also provide a comparison between the efficiency of the proposed algorithm and the spectral flow of Wilson loops.
2501.18291
CueTip: An Interactive and Explainable Physics-aware Pool Assistant
cs.AI cs.HC
We present an interactive and explainable automated coaching assistant called CueTip for a variant of pool/billiards. CueTip's novelty lies in its combination of three features: a natural-language interface, an ability to perform contextual, physics-aware reasoning, and that its explanations are rooted in a set of predetermined guidelines developed by domain experts. We instrument a physics simulator so that it generates event traces in natural language alongside traditional state traces. Event traces lend themselves to interpretation by language models, which serve as the interface to our assistant. We design and train a neural adaptor that decouples tactical choices made by CueTip from its interactivity and explainability allowing it to be reconfigured to mimic any pool playing agent. Our experiments show that CueTip enables contextual query-based assistance and explanations while maintaining the strength of the agent in terms of win rate (improving it in some situations). The explanations generated by CueTip are physically-aware and grounded in the expert rules and are therefore more reliable.
2501.18292
Citation Recommendation based on Argumentative Zoning of User Queries
cs.IR cs.CL cs.DL
Citation recommendation aims to locate the important papers for scholars to cite. When writing the citing sentences, the authors usually hold different citing intents, which are referred to citation function in citation analysis. Since argumentative zoning is to identify the argumentative and rhetorical structure in scientific literature, we want to use this information to improve the citation recommendation task. In this paper, a multi-task learning model is built for citation recommendation and argumentative zoning classification. We also generated an annotated corpus of the data from PubMed Central based on a new argumentative zoning schema. The experimental results show that, by considering the argumentative information in the citing sentence, citation recommendation model will get better performance.
2501.18294
A Comprehensive Analysis on Machine Learning based Methods for Lung Cancer Level Classification
cs.CV cs.AI
Lung cancer is a major issue in worldwide public health, requiring early diagnosis using stable techniques. This work begins a thorough investigation of the use of machine learning (ML) methods for precise classification of lung cancer stages. A cautious analysis is performed to overcome overfitting issues in model performance, taking into account minimum child weight and learning rate. A set of machine learning (ML) models including XGBoost (XGB), LGBM, Adaboost, Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), CatBoost, and k-Nearest Neighbor (k-NN) are run methodically and contrasted. Furthermore, the correlation between features and targets is examined using the deep neural network (DNN) model and thus their capability in detecting complex patternsis established. It is argued that several ML models can be capable of classifying lung cancer stages with great accuracy. In spite of the complexity of DNN architectures, traditional ML models like XGBoost, LGBM, and Logistic Regression excel with superior performance. The models perform better than the others in lung cancer prediction on the complete set of comparative metrics like accuracy, precision, recall, and F-1 score
2501.18296
Extending the design space of ontologization practices: Using bCLEARer as an example
cs.AI
Our aim in this paper is to outline how the design space for the ontologization process is richer than current practice would suggest. We point out that engineering processes as well as products need to be designed - and identify some components of the design. We investigate the possibility of designing a range of radically new practices, providing examples of the new practices from our work over the last three decades with an outlier methodology, bCLEARer. We also suggest that setting an evolutionary context for ontologization helps one to better understand the nature of these new practices and provides the conceptual scaffolding that shapes fertile processes. Where this evolutionary perspective positions digitalization (the evolutionary emergence of computing technologies) as the latest step in a long evolutionary trail of information transitions. This reframes ontologization as a strategic tool for leveraging the emerging opportunities offered by digitalization.
2501.18298
Update Estimation and Scheduling for Over-the-Air Federated Learning with Energy Harvesting Devices
cs.LG cs.DC
We study over-the-air (OTA) federated learning (FL) for energy harvesting devices with heterogeneous data distribution over wireless fading multiple access channel (MAC). To address the impact of low energy arrivals and data heterogeneity on global learning, we propose user scheduling strategies. Specifically, we develop two approaches: 1) entropy-based scheduling for known data distributions and 2) least-squares-based user representation estimation for scheduling with unknown data distributions at the parameter server. Both methods aim to select diverse users, mitigating bias and enhancing convergence. Numerical and analytical results demonstrate improved learning performance by reducing redundancy and conserving energy.
2501.18299
Model-Free RL Agents Demonstrate System 1-Like Intentionality
cs.AI
This paper argues that model-free reinforcement learning (RL) agents, while lacking explicit planning mechanisms, exhibit behaviours that can be analogised to System 1 ("thinking fast") processes in human cognition. Unlike model-based RL agents, which operate akin to System 2 ("thinking slow") reasoning by leveraging internal representations for planning, model-free agents react to environmental stimuli without anticipatory modelling. We propose a novel framework linking the dichotomy of System 1 and System 2 to the distinction between model-free and model-based RL. This framing challenges the prevailing assumption that intentionality and purposeful behaviour require planning, suggesting instead that intentionality can manifest in the structured, reactive behaviours of model-free agents. By drawing on interdisciplinary insights from cognitive psychology, legal theory, and experimental jurisprudence, we explore the implications of this perspective for attributing responsibility and ensuring AI safety. These insights advocate for a broader, contextually informed interpretation of intentionality in RL systems, with implications for their ethical deployment and regulation.
2501.18308
Zero Estimation Cost Strategy for Witsenhausen Counterexample with Causal Encoder
cs.IT math.IT
We propose a zero estimation cost (ZEC) scheme for causal-encoding noncausal-decoding vector-valued Witsenhausen counterexample based on the coordination coding result. In contrast to source coding, our goal is to communicate a controlled system state. The introduced ZEC scheme is a joint control-communication approach that transforms the system state into a sequence that can be efficiently communicated using block coding. Numerical results show that our approach significantly reduces the power required for achieving zero-estimation-cost state reconstruction at the decoder. In the second part, we introduce a more general non-zero estimation cost (Non-ZEC) scheme. We observe numerically that the Non-ZEC scheme operates as a time-sharing mechanism between the two-point strategy and the ZEC scheme. Overall, by leveraging block-coding gain, our proposed methods substantially improve the power-estimation trade-off for Witsenhausen counterexample.
2501.18309
Knowledge in multi-robot systems: an interplay of dynamics, computation and communication
cs.LO cs.DC cs.RO
We show that the hybrid systems perspective of distributed multi-robot systems is compatible with logical models of knowledge already used in distributed computing, and demonstrate its usefulness by deriving sufficient epistemic conditions for exploration and gathering robot tasks to be solvable. We provide a separation of the physical and computational aspects of a robotic system, allowing us to decouple the problems related to each and directly use methods from control theory and distributed computing, fields that are traditionally distant in the literature. Finally, we demonstrate a novel approach for reasoning about the knowledge in multi-robot systems through a principled method of converting a switched hybrid dynamical system into a temporal-epistemic logic model, passing through an abstract state machine representation. This creates space for methods and results to be exchanged across the fields of control theory, distributed computing and temporal-epistemic logic, while reasoning about multi-robot systems.
2501.18310
Efficient Neural Theorem Proving via Fine-grained Proof Structure Analysis
cs.LG cs.AI
The synergy between deep learning models and traditional automation tools plays a pivotal role in developing robust neural theorem provers (NTPs). However, for proof synthesis with LLMs, previous work applies automation tools either only when the model explicitly calls the method, or only at a single granularity level, failing to fully exploit the power of built-in tactics and off-the-shelf automated theorem provers. In this work, we propose ProofAug, a novel theorem proving method that enjoys superior sample efficiency through equipping proof-generation LLMs with automation methods in different granularities via fine-grained structure analysis of model-generated proof proposals. Furthermore, ProofAug serves as a versatile plug-and-play module that seamlessly integrates with any tree-search algorithm, enabling our construction of an efficient recursive proving (ERP) module to further enhance performance. The superiority of our method is validated on the miniF2F-test benchmark using the open-source deepseek-math-7b-base model and the Isabelle proof assistant. Notably, by additionally employing a mixed prompting strategy, we achieve a cumulative pass rate of 66.0% after curation of the dataset (61.9% for the original version), setting a new SOTA across all proof languages with a total sample budget of only 2100. Our code is available at https://github.com/haoxiongliu/ProofAug.
2501.18313
Simulation of microstructures and machine learning
cs.CV
Machine learning offers attractive solutions to challenging image processing tasks. Tedious development and parametrization of algorithmic solutions can be replaced by training a convolutional neural network or a random forest with a high potential to generalize. However, machine learning methods rely on huge amounts of representative image data along with a ground truth, usually obtained by manual annotation. Thus, limited availability of training data is a critical bottleneck. We discuss two use cases: optical quality control in industrial production and segmenting crack structures in 3D images of concrete. For optical quality control, all defect types have to be trained but are typically not evenly represented in the training data. Additionally, manual annotation is costly and often inconsistent. It is nearly impossible in the second case: segmentation of crack systems in 3D images of concrete. Synthetic images, generated based on realizations of stochastic geometry models, offer an elegant way out. A wide variety of structure types can be generated. The within structure variation is naturally captured by the stochastic nature of the models and the ground truth is for free. Many new questions arise. In particular, which characteristics of the real image data have to be met to which degree of fidelity.
2501.18314
AGAV-Rater: Adapting Large Multimodal Model for AI-Generated Audio-Visual Quality Assessment
cs.MM cs.CV cs.SD eess.AS
Many video-to-audio (VTA) methods have been proposed for dubbing silent AI-generated videos. An efficient quality assessment method for AI-generated audio-visual content (AGAV) is crucial for ensuring audio-visual quality. Existing audio-visual quality assessment methods struggle with unique distortions in AGAVs, such as unrealistic and inconsistent elements. To address this, we introduce AGAVQA, the first large-scale AGAV quality assessment dataset, comprising 3,382 AGAVs from 16 VTA methods. AGAVQA includes two subsets: AGAVQA-MOS, which provides multi-dimensional scores for audio quality, content consistency, and overall quality, and AGAVQA-Pair, designed for optimal AGAV pair selection. We further propose AGAV-Rater, a LMM-based model that can score AGAVs, as well as audio and music generated from text, across multiple dimensions, and selects the best AGAV generated by VTA methods to present to the user. AGAV-Rater achieves state-of-the-art performance on AGAVQA, Text-to-Audio, and Text-to-Music datasets. Subjective tests also confirm that AGAV-Rater enhances VTA performance and user experience. The project page is available at https://agav-rater.github.io.
2501.18315
Surface Defect Identification using Bayesian Filtering on a 3D Mesh
cs.CV cs.RO
This paper presents a CAD-based approach for automated surface defect detection. We leverage the a-priori knowledge embedded in a CAD model and integrate it with point cloud data acquired from commercially available stereo and depth cameras. The proposed method first transforms the CAD model into a high-density polygonal mesh, where each vertex represents a state variable in 3D space. Subsequently, a weighted least squares algorithm is employed to iteratively estimate the state of the scanned workpiece based on the captured point cloud measurements. This framework offers the potential to incorporate information from diverse sensors into the CAD domain, facilitating a more comprehensive analysis. Preliminary results demonstrate promising performance, with the algorithm achieving convergence to a sub-millimeter standard deviation in the region of interest using only approximately 50 point cloud samples. This highlights the potential of utilising commercially available stereo cameras for high-precision quality control applications.
2501.18318
Estimating unknown dynamics and cost as a bilinear system with Koopman-based Inverse Optimal Control
eess.SY cs.SY math.DS
In this work, we address the challenge of approximating unknown system dynamics and costs by representing them as a bilinear system using Koopman-based Inverse Optimal Control (IOC). Using optimal trajectories, we construct a bilinear control system in transformed state variables through a modified Extended Dynamic Mode Decomposition with control (EDMDc) that maintains exact dynamical equivalence with the original nonlinear system. We derive Pontryagin's Maximum Principle (PMP) optimality conditions for this system, which closely resemble those of the inverse Linear Quadratic Regulator (LQR) problem due to the consistent control input and state independence from the control. This similarity allows us to apply modified inverse LQR theory, offering a more tractable and robust alternative to nonlinear Inverse Optimal Control methods, especially when dealing with unknown dynamics. Our approach also benefits from the extensive analytical properties of bilinear control systems, providing a solid foundation for further analysis and application. We demonstrate the effectiveness of the proposed method through theoretical analysis, simulation studies and a robotic experiment, highlighting its potential for broader applications in the approximation and design of control systems.
2501.18320
Leveraging LLM Agents for Automated Optimization Modeling for SASP Problems: A Graph-RAG based Approach
cs.AI eess.SP
Automated optimization modeling (AOM) has evoked considerable interest with the rapid evolution of large language models (LLMs). Existing approaches predominantly rely on prompt engineering, utilizing meticulously designed expert response chains or structured guidance. However, prompt-based techniques have failed to perform well in the sensor array signal processing (SASP) area due the lack of specific domain knowledge. To address this issue, we propose an automated modeling approach based on retrieval-augmented generation (RAG) technique, which consists of two principal components: a multi-agent (MA) structure and a graph-based RAG (Graph-RAG) process. The MA structure is tailored for the architectural AOM process, with each agent being designed based on principles of human modeling procedure. The Graph-RAG process serves to match user query with specific SASP modeling knowledge, thereby enhancing the modeling result. Results on ten classical signal processing problems demonstrate that the proposed approach (termed as MAG-RAG) outperforms several AOM benchmarks.
2501.18322
A Unified Perspective on the Dynamics of Deep Transformers
cs.LG math.AP
Transformers, which are state-of-the-art in most machine learning tasks, represent the data as sequences of vectors called tokens. This representation is then exploited by the attention function, which learns dependencies between tokens and is key to the success of Transformers. However, the iterative application of attention across layers induces complex dynamics that remain to be fully understood. To analyze these dynamics, we identify each input sequence with a probability measure and model its evolution as a Vlasov equation called Transformer PDE, whose velocity field is non-linear in the probability measure. Our first set of contributions focuses on compactly supported initial data. We show the Transformer PDE is well-posed and is the mean-field limit of an interacting particle system, thus generalizing and extending previous analysis to several variants of self-attention: multi-head attention, L2 attention, Sinkhorn attention, Sigmoid attention, and masked attention--leveraging a conditional Wasserstein framework. In a second set of contributions, we are the first to study non-compactly supported initial conditions, by focusing on Gaussian initial data. Again for different types of attention, we show that the Transformer PDE preserves the space of Gaussian measures, which allows us to analyze the Gaussian case theoretically and numerically to identify typical behaviors. This Gaussian analysis captures the evolution of data anisotropy through a deep Transformer. In particular, we highlight a clustering phenomenon that parallels previous results in the non-normalized discrete case.
2501.18324
A Video-grounded Dialogue Dataset and Metric for Event-driven Activities
cs.CV cs.CL
This paper presents VDAct, a dataset for a Video-grounded Dialogue on Event-driven Activities, alongside VDEval, a session-based context evaluation metric specially designed for the task. Unlike existing datasets, VDAct includes longer and more complex video sequences that depict a variety of event-driven activities that require advanced contextual understanding for accurate response generation. The dataset comprises 3,000 dialogues with over 30,000 question-and-answer pairs, derived from 1,000 videos with diverse activity scenarios. VDAct displays a notably challenging characteristic due to its broad spectrum of activity scenarios and wide range of question types. Empirical studies on state-of-the-art vision foundation models highlight their limitations in addressing certain question types on our dataset. Furthermore, VDEval, which integrates dialogue session history and video content summaries extracted from our supplementary Knowledge Graphs to evaluate individual responses, demonstrates a significantly higher correlation with human assessments on the VDAct dataset than existing evaluation metrics that rely solely on the context of single dialogue turns.
2501.18328
CodeBrain: Impute Any Brain MRI via Instance-specific Scalar-quantized Codes
cs.CV cs.AI
MRI imputation aims to synthesize the missing modality from one or more available ones, which is highly desirable since it reduces scanning costs and delivers comprehensive MRI information to enhance clinical diagnosis. In this paper, we propose a unified model, CodeBrain, designed to adapt to various brain MRI imputation scenarios. The core design lies in casting various inter-modality transformations as a full-modality code prediction task. To this end, CodeBrain is trained in two stages: Reconstruction and Code Prediction. First, in the Reconstruction stage, we reconstruct each MRI modality, which is mapped into a shared latent space followed by a scalar quantization. Since such quantization is lossy and the code is low dimensional, another MRI modality belonging to the same subject is randomly selected to generate common features to supplement the code and boost the target reconstruction. In the second stage, we train another encoder by a customized grading loss to predict the full-modality codes from randomly masked MRI samples, supervised by the corresponding quantized codes generated from the first stage. In this way, the inter-modality transformation is achieved by mapping the instance-specific codes in a finite scalar space. We evaluated the proposed CodeBrain model on two public brain MRI datasets (i.e., IXI and BraTS 2023). Extensive experiments demonstrate that our CodeBrain model achieves superior imputation performance compared to four existing methods, establishing a new state of the art for unified brain MRI imputation. Codes will be released.
2501.18331
Stream-Based Monitoring of Algorithmic Fairness
cs.LG cs.LO cs.SE
Automatic decision and prediction systems are increasingly deployed in applications where they significantly impact the livelihood of people, such as for predicting the creditworthiness of loan applicants or the recidivism risk of defendants. These applications have given rise to a new class of algorithmic-fairness specifications that require the systems to decide and predict without bias against social groups. Verifying these specifications statically is often out of reach for realistic systems, since the systems may, e.g., employ complex learning components, and reason over a large input space. In this paper, we therefore propose stream-based monitoring as a solution for verifying the algorithmic fairness of decision and prediction systems at runtime. Concretely, we present a principled way to formalize algorithmic fairness over temporal data streams in the specification language RTLola and demonstrate the efficacy of this approach on a number of benchmarks. Besides synthetic scenarios that particularly highlight its efficiency on streams with a scaling amount of data, we notably evaluate the monitor on real-world data from the recidivism prediction tool COMPAS.
2501.18337
Unfaithful Probability Distributions in Binary Triple of Causality Directed Acyclic Graph
stat.ML cs.AI cs.LG
Faithfulness is the foundation of probability distribution and graph in causal discovery and causal inference. In this paper, several unfaithful probability distribution examples are constructed in three--vertices binary causality directed acyclic graph (DAG) structure, which are not faithful to causal DAGs described in J.M.,Robins,et al. Uniform consistency in causal inference. Biometrika (2003),90(3): 491--515. And the general unfaithful probability distribution with multiple independence and conditional independence in binary triple causal DAG is given.
2501.18344
Transfer Learning of Surrogate Models: Integrating Domain Warping and Affine Transformations
cs.LG cs.AI
Surrogate models provide efficient alternatives to computationally demanding real-world processes but often require large datasets for effective training. A promising solution to this limitation is the transfer of pre-trained surrogate models to new tasks. Previous studies have investigated the transfer of differentiable and non-differentiable surrogate models, typically assuming an affine transformation between the source and target functions. This paper extends previous research by addressing a broader range of transformations, including linear and nonlinear variations. Specifically, we consider the combination of an unknown input warping, such as one modelled by the beta cumulative distribution function, with an unspecified affine transformation. Our approach achieves transfer learning by employing a limited number of data points from the target task to optimize these transformations, minimizing empirical loss on the transfer dataset. We validate the proposed method on the widely used Black-Box Optimization Benchmark (BBOB) testbed and a real-world transfer learning task from the automobile industry. The results underscore the significant advantages of the approach, revealing that the transferred surrogate significantly outperforms both the original surrogate and the one built from scratch using the transfer dataset, particularly in data-scarce scenarios.
2501.18350
Joint Power and Spectrum Orchestration for D2D Semantic Communication Underlying Energy-Efficient Cellular Networks
eess.SY cs.SY
Semantic communication (SemCom) has been recently deemed a promising next-generation wireless technique to enable efficient spectrum savings and information exchanges, thus naturally introducing a novel and practical network paradigm where cellular and device-to-device (D2D) SemCom approaches coexist. Nevertheless, the involved wireless resource management becomes complicated and challenging due to the unique semantic performance measurements and energy-consuming semantic coding mechanism. To this end, this paper jointly investigates power control and spectrum reuse problems for energy-efficient D2D SemCom cellular networks. Concretely, we first model the user preference-aware semantic triplet transmission and leverage a novel metric of semantic value to identify the semantic information importance conveyed in SemCom. Then, we define the additional power consumption from semantic encoding in conjunction with basic power amplifier dissipation to derive the overall system energy efficiency (semantics/Joule). Next, we formulate an energy efficiency maximization problem for joint power and spectrum allocation subject to several SemCom-related and practical constraints. Afterward, we propose an optimal resource management solution by employing the fractional-to-subtractive problem transformation and decomposition while developing a three-stage method with theoretical analysis of its optimality guarantee and computational complexity. Numerical results demonstrate the adequate performance superiority of our proposed solution compared with different benchmarks.
2501.18351
Dual-BEV Nav: Dual-layer BEV-based Heuristic Path Planning for Robotic Navigation in Unstructured Outdoor Environments
cs.RO
Path planning with strong environmental adaptability plays a crucial role in robotic navigation in unstructured outdoor environments, especially in the case of low-quality location and map information. The path planning ability of a robot depends on the identification of the traversability of global and local ground areas. In real-world scenarios, the complexity of outdoor open environments makes it difficult for robots to identify the traversability of ground areas that lack a clearly defined structure. Moreover, most existing methods have rarely analyzed the integration of local and global traversability identifications in unstructured outdoor scenarios. To address this problem, we propose a novel method, Dual-BEV Nav, first introducing Bird's Eye View (BEV) representations into local planning to generate high-quality traversable paths. Then, these paths are projected onto the global traversability map generated by the global BEV planning model to obtain the optimal waypoints. By integrating the traversability from both local and global BEV, we establish a dual-layer BEV heuristic planning paradigm, enabling long-distance navigation in unstructured outdoor environments. We test our approach through both public dataset evaluations and real-world robot deployments, yielding promising results. Compared to baselines, the Dual-BEV Nav improved temporal distance prediction accuracy by up to $18.7\%$. In the real-world deployment, under conditions significantly different from the training set and with notable occlusions in the global BEV, the Dual-BEV Nav successfully achieved a 65-meter-long outdoor navigation. Further analysis demonstrates that the local BEV representation significantly enhances the rationality of the planning, while the global BEV probability map ensures the robustness of the overall planning.
2501.18355
Multilayered Intelligent Reflecting Surface for Long-Range Underwater Acoustic Communication
eess.AS cs.SD cs.SY eess.SP eess.SY
This article introduces a multilayered acoustic reconfigurable intelligent surface (ML-ARIS) architecture designed for the next generation of underwater communications. ML-ARIS incorporates multiple layers of piezoelectric material in each acoustic reflector, with the load impedance of each layer independently adjustable via a control circuit. This design increases the flexibility in generating reflected signals with desired amplitudes and orthogonal phases, enabling passive in-phase and quadrature (IQ) modulation using a single acoustic reflector. Such a feature enables precise beam steering, enhancing sound levels in targeted directions while minimizing interference in surrounding environments. Extensive simulations and tank experiments were conducted to verify the feasibility of ML-ARIS. The experimental results indicate that implementing IQ modulation with a multilayer structure is indeed practical in real-world scenarios, making it possible to use a single reflection unit to generate reflected waves with high-resolution amplitudes and phases.
2501.18356
State Stream Transformer (SST) : Emergent Metacognitive Behaviours Through Latent State Persistence
cs.LG cs.AI cs.CL
We introduce the State Stream Transformer (SST), a novel LLM architecture that reveals emergent reasoning behaviours and capabilities latent in pretrained weights through addressing a fundamental limitation in traditional transformer models: the lack of latent computational continuity across autoregressive generations in the state space. SST introduces a sliding window latent state (FFN) cache with weighted decay that maintains and evolves persistent latent processes throughout autoregressive generations. Through controlled experiments comparing base and SST architectures using the same frozen weights, we demonstrate that this architectural modification alone enables enhanced reasoning capabilities which appear best explained by some form of potential higher-order processing, as evidenced by emergent metacognitive behaviours. These behaviours persist under controlled conditions designed to eliminate confounding factors such as stochastic variation or learned response patterns. Analysis of latent state distributions and processing dynamics provides evidence that it is solely the 'state stream' that is responsible for these phenomena. In quantitative evaluations, the SST achieves substantial performance improvements over the base model on two reasoning benchmarks, reaching 89.01\% accuracy on GSM-8K (0-shot) and 91.04\% on ARC Challenge (0-shot CoT). These findings indicate that persistent computation in the latent state space enables fundamentally different information processing and internal reasoning strategies, with implications for our understanding of artificial intelligence systems.
2501.18357
Contrastive Learning Meets Pseudo-label-assisted Mixup Augmentation: A Comprehensive Graph Representation Framework from Local to Global
cs.LG
Graph Neural Networks (GNNs) have demonstrated remarkable effectiveness in various graph representation learning tasks. However, most existing GNNs focus primarily on capturing local information through explicit graph convolution, often neglecting global message-passing. This limitation hinders the establishment of a collaborative interaction between global and local information, which is crucial for comprehensively understanding graph data. To address these challenges, we propose a novel framework called Comprehensive Graph Representation Learning (ComGRL). ComGRL integrates local information into global information to derive powerful representations. It achieves this by implicitly smoothing local information through flexible graph contrastive learning, ensuring reliable representations for subsequent global exploration. Then ComGRL transfers the locally derived representations to a multi-head self-attention module, enhancing their discriminative ability by uncovering diverse and rich global correlations. To further optimize local information dynamically under the self-supervision of pseudo-labels, ComGRL employs a triple sampling strategy to construct mixed node pairs and applies reliable Mixup augmentation across attributes and structure for local contrastive learning. This approach broadens the receptive field and facilitates coordination between local and global representation learning, enabling them to reinforce each other. Experimental results across six widely used graph datasets demonstrate that ComGRL achieves excellent performance in node classification tasks. The code could be available at https://github.com/JinluWang1002/ComGRL.
2501.18359
Contextual Online Decision Making with Infinite-Dimensional Functional Regression
stat.ML cs.LG
Contextual sequential decision-making problems play a crucial role in machine learning, encompassing a wide range of downstream applications such as bandits, sequential hypothesis testing and online risk control. These applications often require different statistical measures, including expectation, variance and quantiles. In this paper, we provide a universal admissible algorithm framework for dealing with all kinds of contextual online decision-making problems that directly learns the whole underlying unknown distribution instead of focusing on individual statistics. This is much more difficult because the dimension of the regression is uncountably infinite, and any existing linear contextual bandits algorithm will result in infinite regret. To overcome this issue, we propose an efficient infinite-dimensional functional regression oracle for contextual cumulative distribution functions (CDFs), where each data point is modeled as a combination of context-dependent CDF basis functions. Our analysis reveals that the decay rate of the eigenvalue sequence of the design integral operator governs the regression error rate and, consequently, the utility regret rate. Specifically, when the eigenvalue sequence exhibits a polynomial decay of order $\frac{1}{\gamma}\ge 1$, the utility regret is bounded by $\tilde{\mathcal{O}}\Big(T^{\frac{3\gamma+2}{2(\gamma+2)}}\Big)$. By setting $\gamma=0$, this recovers the existing optimal regret rate for contextual bandits with finite-dimensional regression and is optimal under a stronger exponential decay assumption. Additionally, we provide a numerical method to compute the eigenvalue sequence of the integral operator, enabling the practical implementation of our framework.
2501.18361
Video-based Surgical Tool-tip and Keypoint Tracking using Multi-frame Context-driven Deep Learning Models
cs.CV
Automated tracking of surgical tool keypoints in robotic surgery videos is an essential task for various downstream use cases such as skill assessment, expertise assessment, and the delineation of safety zones. In recent years, the explosion of deep learning for vision applications has led to many works in surgical instrument segmentation, while lesser focus has been on tracking specific tool keypoints, such as tool tips. In this work, we propose a novel, multi-frame context-driven deep learning framework to localize and track tool keypoints in surgical videos. We train and test our models on the annotated frames from the 2015 EndoVis Challenge dataset, resulting in state-of-the-art performance. By leveraging sophisticated deep learning models and multi-frame context, we achieve 90\% keypoint detection accuracy and a localization RMS error of 5.27 pixels. Results on a self-annotated JIGSAWS dataset with more challenging scenarios also show that the proposed multi-frame models can accurately track tool-tip and tool-base keypoints, with ${<}4.2$-pixel RMS error overall. Such a framework paves the way for accurately tracking surgical instrument keypoints, enabling further downstream use cases. Project and dataset webpage: https://tinyurl.com/mfc-tracker
2501.18362
MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding
cs.AI cs.CL cs.CV cs.LG
We introduce MedXpertQA, a highly challenging and comprehensive benchmark to evaluate expert-level medical knowledge and advanced reasoning. MedXpertQA includes 4,460 questions spanning 17 specialties and 11 body systems. It includes two subsets, Text for text evaluation and MM for multimodal evaluation. Notably, MM introduces expert-level exam questions with diverse images and rich clinical information, including patient records and examination results, setting it apart from traditional medical multimodal benchmarks with simple QA pairs generated from image captions. MedXpertQA applies rigorous filtering and augmentation to address the insufficient difficulty of existing benchmarks like MedQA, and incorporates specialty board questions to improve clinical relevance and comprehensiveness. We perform data synthesis to mitigate data leakage risk and conduct multiple rounds of expert reviews to ensure accuracy and reliability. We evaluate 16 leading models on MedXpertQA. Moreover, medicine is deeply connected to real-world decision-making, providing a rich and representative setting for assessing reasoning abilities beyond mathematics and code. To this end, we develop a reasoning-oriented subset to facilitate the assessment of o1-like models.
2501.18363
Robust Online Conformal Prediction under Uniform Label Noise
cs.LG
Conformal prediction is an emerging technique for uncertainty quantification that constructs prediction sets guaranteed to contain the true label with a predefined probability. Recent work develops online conformal prediction methods that adaptively construct prediction sets to accommodate distribution shifts. However, existing algorithms typically assume perfect label accuracy which rarely holds in practice. In this work, we investigate the robustness of online conformal prediction under uniform label noise with a known noise rate, in both constant and dynamic learning rate schedules. We show that label noise causes a persistent gap between the actual mis-coverage rate and the desired rate $\alpha$, leading to either overestimated or underestimated coverage guarantees. To address this issue, we propose Noise Robust Online Conformal Prediction (dubbed NR-OCP) by updating the threshold with a novel robust pinball loss, which provides an unbiased estimate of clean pinball loss without requiring ground-truth labels. Our theoretical analysis shows that NR-OCP eliminates the coverage gap in both constant and dynamic learning rate schedules, achieving a convergence rate of $\mathcal{O}(T^{-1/2})$ for both empirical and expected coverage errors under uniform label noise. Extensive experiments demonstrate the effectiveness of our method by achieving both precise coverage and improved efficiency.
2501.18365
RbFT: Robust Fine-tuning for Retrieval-Augmented Generation against Retrieval Defects
cs.CL cs.IR
Retrieval-augmented generation (RAG) enhances large language models (LLMs) by integrating external knowledge retrieved from a knowledge base. However, its effectiveness is fundamentally constrained by the reliability of both the retriever and the knowledge base. In real-world scenarios, imperfections in these components often lead to the retrieval of noisy, irrelevant, or misleading counterfactual information, ultimately undermining the trustworthiness of RAG systems. To address this challenge, we propose Robust Fine-Tuning (RbFT), a method designed to enhance the resilience of LLMs against retrieval defects through two targeted fine-tuning tasks. Experimental results demonstrate that RbFT significantly improves the robustness of RAG systems across diverse retrieval conditions, surpassing existing methods while maintaining high inference efficiency and compatibility with other robustness techniques.
2501.18367
A Learnable Multi-views Contrastive Framework with Reconstruction Discrepancy for Medical Time-Series
cs.LG cs.AI
In medical time series disease diagnosis, two key challenges are identified.First, the high annotation cost of medical data leads to overfitting in models trained on label-limited, single-center datasets. To address this, we propose incorporating external data from related tasks and leveraging AE-GAN to extract prior knowledge,providing valuable references for downstream tasks. Second, many existing studies employ contrastive learning to derive more generalized medical sequence representations for diagnostic tasks, usually relying on manually designed diverse positive and negative sample pairs.However, these approaches are complex, lack generalizability, and fail to adaptively capture disease-specific features across different conditions.To overcome this, we introduce LMCF (Learnable Multi-views Contrastive Framework), a framework that integrates a multi-head attention mechanism and adaptively learns representations from different views through inter-view and intra-view contrastive learning strategies.Additionally, the pre-trained AE-GAN is used to reconstruct discrepancies in the target data as disease probabilities, which are then integrated into the contrastive learning process.Experiments on three target datasets demonstrate that our method consistently outperforms seven other baselines, highlighting its significant impact on healthcare applications such as the diagnosis of myocardial infarction, Alzheimer's disease, and Parkinson's disease.
2501.18369
A Cartesian Encoding Graph Neural Network for Crystal Structures Property Prediction: Application to Thermal Ellipsoid Estimation
cs.LG
In diffraction-based crystal structure analysis, thermal ellipsoids, quantified via Anisotropic Displacement Parameters (ADPs), are critical yet challenging to determine. ADPs capture atomic vibrations, reflecting thermal and structural properties, but traditional computation is often expensive. This paper introduces CartNet, a novel graph neural network (GNN) for efficiently predicting crystal properties by encoding atomic geometry into Cartesian coordinates alongside the crystal temperature. CartNet integrates a neighbour equalization technique to emphasize covalent and contact interactions, and a Cholesky-based head to ensure valid ADP predictions. We also propose a rotational SO(3) data augmentation strategy during training to handle unseen orientations. An ADP dataset with over 200,000 experimental crystal structures from the Cambridge Structural Database (CSD) was curated to validate the approach. CartNet significantly reduces computational costs and outperforms existing methods in ADP prediction by 10.87%, while delivering a 34.77% improvement over theoretical approaches. We further evaluated CartNet on other datasets covering formation energy, band gap, total energy, energy above the convex hull, bulk moduli, and shear moduli, achieving 7.71% better results on the Jarvis Dataset and 13.16% on the Materials Project Dataset. These gains establish CartNet as a state-of-the-art solution for diverse crystal property predictions. Project website and online demo: https://www.ee.ub.edu/cartnet
2501.18373
Function Encoders: A Principled Approach to Transfer Learning in Hilbert Spaces
cs.LG
A central challenge in transfer learning is designing algorithms that can quickly adapt and generalize to new tasks without retraining. Yet, the conditions of when and how algorithms can effectively transfer to new tasks is poorly characterized. We introduce a geometric characterization of transfer in Hilbert spaces and define three types of inductive transfer: interpolation within the convex hull, extrapolation to the linear span, and extrapolation outside the span. We propose a method grounded in the theory of function encoders to achieve all three types of transfer. Specifically, we introduce a novel training scheme for function encoders using least-squares optimization, prove a universal approximation theorem for function encoders, and provide a comprehensive comparison with existing approaches such as transformers and meta-learning on four diverse benchmarks. Our experiments demonstrate that the function encoder outperforms state-of-the-art methods on four benchmark tasks and on all three types of transfer.
2501.18374
Proofs for Folklore Theorems on the Radon-Nikodym Derivative
cs.IT math.HO math.IT math.ST stat.ML stat.TH
Rigorous statements and formal proofs are presented for both foundational and advanced folklore theorems on the Radon-Nikodym derivative. The cases of product and marginal measures are carefully considered; and the hypothesis under which the statements hold are rigorously enumerated.
2501.18376
Cracks in concrete
cs.CV eess.IV stat.AP
Finding and properly segmenting cracks in images of concrete is a challenging task. Cracks are thin and rough and being air filled do yield a very weak contrast in 3D images obtained by computed tomography. Enhancing and segmenting dark lower-dimensional structures is already demanding. The heterogeneous concrete matrix and the size of the images further increase the complexity. ML methods have proven to solve difficult segmentation problems when trained on enough and well annotated data. However, so far, there is not much 3D image data of cracks available at all, let alone annotated. Interactive annotation is error-prone as humans can easily tell cats from dogs or roads without from roads with cars but have a hard time deciding whether a thin and dark structure seen in a 2D slice continues in the next one. Training networks by synthetic, simulated images is an elegant way out, bears however its own challenges. In this contribution, we describe how to generate semi-synthetic image data to train CNN like the well known 3D U-Net or random forests for segmenting cracks in 3D images of concrete. The thickness of real cracks varies widely, both, within one crack as well as from crack to crack in the same sample. The segmentation method should therefore be invariant with respect to scale changes. We introduce the so-called RieszNet, designed for exactly this purpose. Finally, we discuss how to generalize the ML crack segmentation methods to other concrete types.
2501.18377
Using Read Promotion and Mixed Isolation Levels for Performant Yet Serializable Execution of Transaction Programs
cs.DB
We propose a theory that can determine the lowest isolation level that can be allocated to each transaction program in an application in a mixed-isolation-level setting, to guarantee that all executions will be serializable and thus preserve all integrity constraints, even those that are not explicitly declared. This extends prior work applied to completely known transactions, to deal with the realistic situation where transactions are generated by running programs with parameters that are not known in advance. Using our theory, we propose an optimization method that allows for high throughput while ensuring that all executions are serializable. Our method is based on searching for application code modifications that are semantics-preserving while improving the isolation level allocation. We illustrate our approach to the SmallBank benchmark.
2501.18381
Implicit Riemannian Optimism with Applications to Min-Max Problems
math.OC cs.LG
We introduce a Riemannian optimistic online learning algorithm for Hadamard manifolds based on inexact implicit updates. Unlike prior work, our method can handle in-manifold constraints, and matches the best known regret bounds in the Euclidean setting with no dependence on geometric constants, like the minimum curvature. Building on this, we develop algorithms for g-convex, g-concave smooth min-max problems on Hadamard manifolds. Notably, one method nearly matches the gradient oracle complexity of the lower bound for Euclidean problems, for the first time.
2501.18382
Rydberg Atomic Quantum Receivers for the Multi-User MIMO Uplink
eess.SP cs.IT math.IT
Rydberg atomic quantum receivers exhibit great potential in assisting classical wireless communications due to their outstanding advantages in detecting radio frequency signals. To realize this potential, we integrate a Rydberg atomic quantum receiver into a classical multi-user multiple-input multiple-output (MIMO) scheme to form a multi-user Rydberg atomic quantum MIMO (RAQ-MIMO) system for the uplink. To study this system, we first construct an equivalent baseband signal model, which facilitates convenient system design, signal processing and optimizations. We then study the ergodic achievable rates under both the maximum ratio combining (MRC) and zero-forcing (ZF) schemes by deriving their tight lower bounds. We next compare the ergodic achievable rates of the RAQ-MIMO and the conventional massive MIMO schemes by offering a closed-form expression for the difference of their ergodic achievable rates, which allows us to directly compare the two systems. Our results show that RAQ-MIMO allows the average transmit power of users to be $\sim 20$ dBm lower than that of the conventional massive MIMO. Viewed from a different perspective, an extra $\sim 7$ bits/s/Hz/user rate becomes achievable by ZF RAQ-MIMO, when equipping $50 \sim 500$ receive elements for receiving $1 \sim 100$ user signals at an enough transmit power (e.g., $\ge 20$ dBm).
2501.18385
Performance guarantees for optimization-based state estimation using turnpike properties
math.OC cs.SY eess.SY
In this paper, we develop novel accuracy and performance guarantees for optimal state estimation of general nonlinear systems (in particular, moving horizon estimation, MHE). Our results rely on a turnpike property of the optimal state estimation problem, which essentially states that the omniscient infinite-horizon solution involving all past and future data serves as turnpike for the solutions of finite-horizon estimation problems involving a subset of the data. This leads to the surprising observation that MHE problems naturally exhibit a leaving arc, which may have a strong negative impact on the estimation accuracy. To address this, we propose a delayed MHE scheme, and we show that the resulting performance (both averaged and non-averaged) is approximately optimal and achieves bounded dynamic regret with respect to the infinite-horizon solution, with error terms that can be made arbitrarily small by an appropriate choice of the delay. In various simulation examples, we observe that already a very small delay in the MHE scheme is sufficient to significantly improve the overall estimation error by 20-25 % compared to standard MHE (without delay). This finding is of great importance for practical applications (especially for monitoring, fault detection, and parameter estimation) where a small delay in the estimation is rather irrelevant but may significantly improve the estimation results.
2501.18388
Improved Replicable Boosting with Majority-of-Majorities
cs.LG
We introduce a new replicable boosting algorithm which significantly improves the sample complexity compared to previous algorithms. The algorithm works by doing two layers of majority voting, using an improved version of the replicable boosting algorithm introduced by Impagliazzo et al. [2022] in the bottom layer.
2501.18394
Quantum-Key Distribution using Decoy Pulses to Combat Photon-Number Splitting by Eavesdropper: An Event-by-Event Impairment Enumeration Approach for Performance Evaluation and Design
quant-ph cs.CR cs.IT cs.NI math.IT
Quantum-key distribution (QKD) schemes employing quantum communication links are typically based on the transmission of weak optical pulses over optical fibers to setup a secret key between the transmitting and receiving nodes. Alice transmits optically a random bit stream to the receiver (Bob) through the photon polarizations or the quadrature components of the lightwaves associated with the photons, with a secret key remaining implicitly embedded therein. However, during the above transmission, some eavesdropper (Eve) might attempt to tap the passing-by photons from the optical fiber links to extract the key. In one of the popular QKD schemes, along with signal pulses, some additional decoy pulses are transmitted by Alice, while Eve might use photon-number splitting (PNS) for eavesdropping. In a typical PNS scheme, (i) the optical pulses with single photon are blocked by Eve, (ii) from the optical pulses with two photons, one photon is retained by Eve to carry out eavesdropping operation and the other is retransmitted to Bob, and (iii) all other pulses with more than two photons are retransmitted by Eve to Bob without retaining any photon from them. Extensive theoretical research has been carried out on such QKD schemes, by employing information-theoretic approach along with computer simulations and experimental studies. We present a novel event-by-event impairment enumeration approach to evaluate the overall performance of one such QKD scheme analytically with due consideration to the physical layer of the quantum communication links. The proposed approach monitors the impairments of the propagating optical pulses event-by-event at all possible locations along the optical fiber link using statistical approach, and provides estimates of the realizable key generation rate, while assuring an adequate yield ratio between signal and decoy pulses for the detection of possible eavesdropping.
2501.18397
A weakly compressible SPH method for RANS simulation of wall-bounded turbulent flows
physics.flu-dyn cs.CE
This paper presents a Weakly Compressible Smoothed Particle Hydrodynamics (WCSPH) method for solving the two-equation Reynolds-Averaged Navier-Stokes (RANS) model. The turbulent wall-bounded flow with or without mild flow separation, a crucial flow pattern in engineering applications, yet rarely explored in the SPH community, is simulated. The inconsistency between the Lagrangian characteristic and RANS model, mainly due to the intense particle shear and near-wall discontinuity, is firstly revealed and addressed by the mainstream and nearwall improvements, respectively. The mainstream improvements, including Adaptive Riemann-eddy Dissipation (ARD) and Limited Transport Velocity Formulation (LTVF), address dissipation incompatibility and turbulent kinetic energy over-prediction issues. The nearwall improvements, such as the particle-based wall model realization, weighted near-wall compensation scheme, and constant $y_p$ strategy, improve the accuracy and stability of the adopted wall model, where the wall dummy particles are still used for future coupling of solid dynamics. Besides, to perform rigorous convergence tests, an level-set-based boundary-offset technique is developed to ensure consistent $y^+$ across different resolutions. The benchmark wall-bounded turbulent cases, including straight, mildly- and strongly-curved, and Half Converging and Diverging (HCD) channels are calculated. Good convergence is, to our best knowledge, firstly achieved for both velocity and turbulent kinetic energy for the SPH-RANS method. All the results agree well with the data from the experiments or simulated by the Eulerian methods at engineering-acceptable resolutions. The proposed method bridges particle-based and mesh-based RANS models, providing adaptability for other turbulence models and potential for turbulent fluid-structure interaction (FSI) simulations.
2501.18401
MatIR: A Hybrid Mamba-Transformer Image Restoration Model
cs.CV
In recent years, Transformers-based models have made significant progress in the field of image restoration by leveraging their inherent ability to capture complex contextual features. Recently, Mamba models have made a splash in the field of computer vision due to their ability to handle long-range dependencies and their significant computational efficiency compared to Transformers. However, Mamba currently lags behind Transformers in contextual learning capabilities. To overcome the limitations of these two models, we propose a Mamba-Transformer hybrid image restoration model called MatIR. Specifically, MatIR cross-cycles the blocks of the Transformer layer and the Mamba layer to extract features, thereby taking full advantage of the advantages of the two architectures. In the Mamba module, we introduce the Image Inpainting State Space (IRSS) module, which traverses along four scan paths to achieve efficient processing of long sequence data. In the Transformer module, we combine triangular window-based local attention with channel-based global attention to effectively activate the attention mechanism over a wider range of image pixels. Extensive experimental results and ablation studies demonstrate the effectiveness of our approach.
2501.18403
Efficient Transformer for High Resolution Image Motion Deblurring
cs.CV cs.AI
This paper presents a comprehensive study and improvement of the Restormer architecture for high-resolution image motion deblurring. We introduce architectural modifications that reduce model complexity by 18.4% while maintaining or improving performance through optimized attention mechanisms. Our enhanced training pipeline incorporates additional transformations including color jitter, Gaussian blur, and perspective transforms to improve model robustness as well as a new frequency loss term. Extensive experiments on the RealBlur-R, RealBlur-J, and Ultra-High-Definition Motion blurred (UHDM) datasets demonstrate the effectiveness of our approach. The improved architecture shows better convergence behavior and reduced training time while maintaining competitive performance across challenging scenarios. We also provide detailed ablation studies analyzing the impact of our modifications on model behavior and performance. Our results suggest that thoughtful architectural simplification combined with enhanced training strategies can yield more efficient yet equally capable models for motion deblurring tasks. Code and Data Available at: https://github.com/hamzafer/image-deblurring
2501.18405
Segmentation of cracks in 3d images of fiber reinforced concrete using deep learning
cs.LG
Cracks in concrete structures are very common and are an integral part of this heterogeneous material. Characteristics of cracks induced by standardized tests yield valuable information about the tested concrete formulation and its mechanical properties. Observing cracks on the surface of the concrete structure leaves a wealth of structural information unused. Computed tomography enables looking into the sample without interfering or destroying the microstructure. The reconstructed tomographic images are 3d images, consisting of voxels whose gray values represent local X-ray absorption. In order to identify voxels belonging to the crack, so to segment the crack structure in the images, appropriate algorithms need to be developed. Convolutional neural networks are known to solve this type of task very well given enough and consistent training data. We adapted a 3d version of the well-known U-Net and trained it on semi-synthetic 3d images of real concrete samples equipped with simulated crack structures. Here, we explain the general approach. Moreover, we show how to teach the network to detect also real crack systems in 3d images of varying types of real concrete, in particular of fiber reinforced concrete.
2501.18407
Degree is Important: On Evolving Homogeneous Boolean Functions
cs.NE cs.CR
Boolean functions with good cryptographic properties like high nonlinearity and algebraic degree play an important in the security of stream and block ciphers. Such functions may be designed, for instance, by algebraic constructions or metaheuristics. This paper investigates the use of Evolutionary Algorithms (EAs) to design homogeneous bent Boolean functions, i.e., functions that are maximally nonlinear and whose algebraic normal form contains only monomials of the same degree. In our work, we evaluate three genotype encodings and four fitness functions. Our results show that while EAs manage to find quadratic homogeneous bent functions (with the best method being a GA leveraging a restricted encoding), none of the approaches result in cubic homogeneous bent functions.
2501.18411
Gravity-Bench-v1: A Benchmark on Gravitational Physics Discovery for Agents
cs.AI astro-ph.IM physics.comp-ph
Modern science emerged from reasoning over repeatedly-observed planetary motions. We present Gravity-Bench-v1, an environment-based benchmark that challenges AI agents on tasks that parallel this historical development. Gravity-Bench-v1 evaluates agents on the discovery of physics concealed within a dynamic environment, using rigorous gravitational dynamics simulations. Gravity-Bench includes out-of-distribution cases, i.e. with physics that deviates from the real world, to evaluate true scientific generalization capabilities. Agents must plan to collect data within an experimental budget and must perform a dynamic form of data analysis and reasoning to solve tasks efficiently. Our benchmark admits an open-ended space of solutions. PhD-level solutions for each task are provided, to calibrate AI performance against human expertise. Technically at an upper-undergraduate level, our benchmark proves challenging to baseline AI agents. Gravity-Bench-v1 and planned extensions should help map out AI progress towards scientific discovery capabilities.
2501.18412
Real Time Scheduling Framework for Multi Object Detection via Spiking Neural Networks
eess.SY cs.CV cs.NE cs.SY
Given the energy constraints in autonomous mobile agents (AMAs), such as unmanned vehicles, spiking neural networks (SNNs) are increasingly favored as a more efficient alternative to traditional artificial neural networks. AMAs employ multi-object detection (MOD) from multiple cameras to identify nearby objects while ensuring two essential objectives, (R1) timing guarantee and (R2) high accuracy for safety. In this paper, we propose RT-SNN, the first system design, aiming at achieving R1 and R2 in SNN-based MOD systems on AMAs. Leveraging the characteristic that SNNs gather feature data of input image termed as membrane potential, through iterative computation over multiple timesteps, RT-SNN provides multiple execution options with adjustable timesteps and a novel method for reusing membrane potential to support R1. Then, it captures how these execution strategies influence R2 by introducing a novel notion of mean absolute error and membrane confidence. Further, RT-SNN develops a new scheduling framework consisting of offline schedulability analysis for R1 and a run-time scheduling algorithm for R2 using the notion of membrane confidence. We deployed RT-SNN to Spiking-YOLO, the SNN-based MOD model derived from ANN-to-SNN conversion, and our experimental evaluation confirms its effectiveness in meeting the R1 and R2 requirements while providing significant energy efficiency.
2501.18413
GBFRS: Robust Fuzzy Rough Sets via Granular-ball Computing
cs.AI cs.LG
Fuzzy rough set theory is effective for processing datasets with complex attributes, supported by a solid mathematical foundation and closely linked to kernel methods in machine learning. Attribute reduction algorithms and classifiers based on fuzzy rough set theory exhibit promising performance in the analysis of high-dimensional multivariate complex data. However, most existing models operate at the finest granularity, rendering them inefficient and sensitive to noise, especially for high-dimensional big data. Thus, enhancing the robustness of fuzzy rough set models is crucial for effective feature selection. Muiti-garanularty granular-ball computing, a recent development, uses granular-balls of different sizes to adaptively represent and cover the sample space, performing learning based on these granular-balls. This paper proposes integrating multi-granularity granular-ball computing into fuzzy rough set theory, using granular-balls to replace sample points. The coarse-grained characteristics of granular-balls make the model more robust. Additionally, we propose a new method for generating granular-balls, scalable to the entire supervised method based on granular-ball computing. A forward search algorithm is used to select feature sequences by defining the correlation between features and categories through dependence functions. Experiments demonstrate the proposed model's effectiveness and superiority over baseline methods.
2501.18415
Consensus statement on the credibility assessment of ML predictors
q-bio.QM cs.LG
The rapid integration of machine learning (ML) predictors into in silico medicine has revolutionized the estimation of quantities of interest (QIs) that are otherwise challenging to measure directly. However, the credibility of these predictors is critical, especially when they inform high-stakes healthcare decisions. This position paper presents a consensus statement developed by experts within the In Silico World Community of Practice. We outline twelve key statements forming the theoretical foundation for evaluating the credibility of ML predictors, emphasizing the necessity of causal knowledge, rigorous error quantification, and robustness to biases. By comparing ML predictors with biophysical models, we highlight unique challenges associated with implicit causal knowledge and propose strategies to ensure reliability and applicability. Our recommendations aim to guide researchers, developers, and regulators in the rigorous assessment and deployment of ML predictors in clinical and biomedical contexts.
2501.18416
Exploring Potential Prompt Injection Attacks in Federated Military LLMs and Their Mitigation
cs.LG
Federated Learning (FL) is increasingly being adopted in military collaborations to develop Large Language Models (LLMs) while preserving data sovereignty. However, prompt injection attacks-malicious manipulations of input prompts-pose new threats that may undermine operational security, disrupt decision-making, and erode trust among allies. This perspective paper highlights four potential vulnerabilities in federated military LLMs: secret data leakage, free-rider exploitation, system disruption, and misinformation spread. To address these potential risks, we propose a human-AI collaborative framework that introduces both technical and policy countermeasures. On the technical side, our framework uses red/blue team wargaming and quality assurance to detect and mitigate adversarial behaviors of shared LLM weights. On the policy side, it promotes joint AI-human policy development and verification of security protocols. Our findings will guide future research and emphasize proactive strategies for emerging military contexts.
2501.18417
Real-Time Anomaly Detection with Synthetic Anomaly Monitoring (SAM)
cs.LG
Anomaly detection is essential for identifying rare and significant events across diverse domains such as finance, cybersecurity, and network monitoring. This paper presents Synthetic Anomaly Monitoring (SAM), an innovative approach that applies synthetic control methods from causal inference to improve both the accuracy and interpretability of anomaly detection processes. By modeling normal behavior through the treatment of each feature as a control unit, SAM identifies anomalies as deviations within this causal framework. We conducted extensive experiments comparing SAM with established benchmark models, including Isolation Forest, Local Outlier Factor (LOF), k-Nearest Neighbors (kNN), and One-Class Support Vector Machine (SVM), across five diverse datasets, including Credit Card Fraud, HTTP Dataset CSIC 2010, and KDD Cup 1999, among others. Our results demonstrate that SAM consistently delivers robust performance, highlighting its potential as a powerful tool for real-time anomaly detection in dynamic and complex environments.
2501.18418
Task-based Regularization in Penalized Least-Squares for Binary Signal Detection Tasks in Medical Image Denoising
eess.IV cs.CV
Image denoising algorithms have been extensively investigated for medical imaging. To perform image denoising, penalized least-squares (PLS) problems can be designed and solved, in which the penalty term encodes prior knowledge of the object being imaged. Sparsity-promoting penalties, such as total variation (TV), have been a popular choice for regularizing image denoising problems. However, such hand-crafted penalties may not be able to preserve task-relevant information in measured image data and can lead to oversmoothed image appearances and patchy artifacts that degrade signal detectability. Supervised learning methods that employ convolutional neural networks (CNNs) have emerged as a popular approach to denoising medical images. However, studies have shown that CNNs trained with loss functions based on traditional image quality measures can lead to a loss of task-relevant information in images. Some previous works have investigated task-based loss functions that employ model observers for training the CNN denoising models. However, such training processes typically require a large number of noisy and ground-truth (noise-free or low-noise) image data pairs. In this work, we propose a task-based regularization strategy for use with PLS in medical image denoising. The proposed task-based regularization is associated with the likelihood of linear test statistics of noisy images for Gaussian noise models. The proposed method does not require ground-truth image data and solves an individual optimization problem for denoising each image. Computer-simulation studies are conducted that consider a multivariate-normally distributed (MVN) lumpy background and a binary texture background. It is demonstrated that the proposed regularization strategy can effectively improve signal detectability in denoised images.
2501.18423
DeepExtractor: Time-domain reconstruction of signals and glitches in gravitational wave data with deep learning
gr-qc astro-ph.IM cs.LG physics.data-an physics.ins-det
Gravitational wave (GW) interferometers, detect faint signals from distant astrophysical events, such as binary black hole mergers. However, their high sensitivity also makes them susceptible to background noise, which can obscure these signals. This noise often includes transient artifacts called "glitches" that can mimic astrophysical signals or mask their characteristics. Fast and accurate reconstruction of both signals and glitches is crucial for reliable scientific inference. In this study, we present DeepExtractor, a deep learning framework designed to reconstruct signals and glitches with power exceeding interferometer noise, regardless of their source. We design DeepExtractor to model the inherent noise distribution of GW interferometers, following conventional assumptions that the noise is Gaussian and stationary over short time scales. It operates by predicting and subtracting the noise component of the data, retaining only the clean reconstruction. Our approach achieves superior generalization capabilities for arbitrary signals and glitches compared to methods that directly map inputs to the clean training waveforms. We validate DeepExtractor's effectiveness through three experiments: (1) reconstructing simulated glitches injected into simulated detector noise, (2) comparing performance with the state-of-the-art BayesWave algorithm, and (3) analyzing real data from the Gravity Spy dataset to demonstrate effective glitch subtraction from LIGO strain data. DeepExtractor achieves a median mismatch of only 0.9% for simulated glitches, outperforming several deep learning baselines. Additionally, DeepExtractor surpasses BayesWave in glitch recovery, offering a dramatic computational speedup by reconstructing one glitch sample in approx. 0.1 seconds on a CPU, compared to BayesWave's processing time of approx. one hour per glitch.
2501.18426
Guaranteed confidence-band enclosures for PDE surrogates
cs.LG cs.AI
We propose a method for obtaining statistically guaranteed confidence bands for functional machine learning techniques: surrogate models which map between function spaces, motivated by the need build reliable PDE emulators. The method constructs nested confidence sets on a low-dimensional representation (an SVD) of the surrogate model's prediction error, and then maps these sets to the prediction space using set-propagation techniques. The result are conformal-like coverage guaranteed prediction sets for functional surrogate models. We use zonotopes as basis of the set construction, due to their well studied set-propagation and verification properties. The method is model agnostic and can thus be applied to complex Sci-ML models, including Neural Operators, but also in simpler settings. We also elicit a technique to capture the truncation error of the SVD, ensuring the guarantees of the method.