Machine Learning
Automated Optimization Modeling via a Localizable Error-Driven Perspective
Weiting Liu, Han Wu, Yufei Kuang et al. (+4 more)
Automated optimization modeling via Large
Language Models (LLMs) has emerged as a promising approach to assist complex human decision-making. While post-training has become a pivotal technique to enhance LLMs' capabilities in this domain, its effectiveness is severely constrained by the scarcity and underutilization of high-quality training data. However, through a detailed profiling of error patterns across various problem-response pairs drawn from post-training, we identify two fundamental limitations of existing automated optimization modeling approaches: (L1) the sparsity of error-specific problems and (L2) the sparse rewards associated with difficult problems. We demonstrate that these limitations can result in suboptimal performance in domain-specific post-training for LLMs. To tackle the above two limitations, we propose a novel error-driven learning framework – namely, automated optimization modeling via a localizable error-driven perspective (MIND) – that customizes the whole model training framework from data synthesis to post-training. MIND is based on our key observation of the unique localizable patterns in error propagation of optimization modelings, that is, modeling errors may remain localized to specific semantic segments and do not propagate throughout the entire solution. Thus, in contrast to holistic
reasoning tasks such as mathematical proofs, MIND leverages the construction of a focused, high-density training corpus and proposes Dynamic Supervised Fine-Tuning Policy Optimization (DFPO) to tackle difficult problems through localized refinement. Experiments on six
benchmarks demonstrate that MIND consistently outperforms all the state-of-the-art automated optimization modeling approaches.
Read the full abstract ▾
All authors Weiting Liu, Han Wu, Yufei Kuang, Xiongwei Han, Tao Zhong, Jianfeng Feng, Wenlian Lu
Machine Learning
ELLMPEG: An Edge-based Agentic LLM Video Processing Tool
Zoha Azimi, Reza Farahani, Radu Prodan et al. (+1 more)
Large
language models (LLMs), the foundation of generative AI systems like ChatGPT, are transforming many fields and applications, including multimedia, enabling more advanced content generation, analysis, and interaction. However, cloud-based LLM deployments face three key limitations: high computational and energy demands,
privacy and reliability risks from remote processing, and recurring API costs. Recent advances in agentic AI, especially in structured
reasoning and tool use, offer a better way to exploit open and locally deployed tools and LLMs. This paper presents ELLMPEG, an edge-enabled agentic LLM framework for the automated generation of
video-processing commands. ELLMPEG integrates tool-aware
Retrieval-Augmented Generation (RAG) with iterative self-reflection to produce and locally verify executable FFmpeg and VVenC commands directly at the edge, eliminating reliance on external cloud APIs. To evaluate ELLMPEG, we collect a dedicated
prompt dataset comprising 480 diverse queries covering different categories of FFmpeg and the Versatile
Video Codec (VVC) encoder (VVenC) commands. We validate command generation accuracy and evaluate four open-source LLMs based on command validity,
tokens generated per second, inference time, and energy efficiency. We also execute the generated commands to assess their runtime correctness and practical applicability. Experimental results show that Qwen2.5, when augmented with the ELLMPEG framework, achieves an average command-generation accuracy of 78 % with zero recurring API cost, outperforming all other open-source models across both the FFmpeg and VVenC
datasets.
Read the full abstract ▾
All authors Zoha Azimi, Reza Farahani, Radu Prodan, Christian Timmerer
Machine Learning
Representation Learning Enhanced Deep Reinforcement Learning for Optimal Operation of Hydrogen-based Multi-Energy Systems
Zhenyu Pu, Yu Yang, Lun Yang et al. (+3 more)
Hydrogen-based multi-energy systems (HMES) have emerged as a promising low-carbon and energy-efficient solution, as it can enable the coordinated operation of electricity, heating and cooling supply and demand to enhance operational flexibility, improve overall energy efficiency, and increase the share of renewable integration. However, the optimal operation of HMES remains challenging due to the nonlinear and multi-physics coupled dynamics of hydrogen energy storage systems (HESS) (consisting of electrolyters, fuel
cells and hydrogen tanks) as well as the presence of multiple uncertainties from supply and demand. To address these challenges, this paper develops a comprehensive operational model for HMES that fully captures the nonlinear dynamics and multi-physics process of HESS. Moreover, we propose an enhanced deep reinforcement learning (DRL) framework by integrating the emerging representation learning techniques, enabling substantially accelerated and improved policy optimization for spatially and temporally coupled complex networked systems, which is not provided by conventional DRL. Experimental studies based on real-world
datasets show that the comprehensive model is crucial to ensure the safe and reliable of HESS. In addition, the proposed SR-DRL approaches demonstrate superior convergence rate and performance over conventional DRL counterparts in terms of reducing the operation cost of HMES and handling the system operating constraints. Finally, we provide some insights into the role of representation learning in DRL, speculating that it can reorganize the original state space into a well-structured and cluster-aware
geometric representation, thereby smoothing and facilitating the learning process of DRL.
Read the full abstract ▾
All authors Zhenyu Pu, Yu Yang, Lun Yang, Qing-Shan Jia, Xiaohong Guan, Costas J. Spanos
Machine Learning
Low-Cost Hard-Label Adversarial Attack with Theoretical Foundations
Jun Liu, Leo Yu Zhang, Fengpeng Li et al. (+2 more)
Hard-label black-box attacks, relying solely on top-1 predictions, represent one of the most challenging yet practically threat models. Despite recent progress, existing approaches face two key limitations: (1) they overlook the critical role of initialization, focusing primarily on optimization strategies; and (2) they rely heavily on empirical heuristics without theoretical guarantees. To bridge this gap, we establish a unified theoretical framework showing that existing sign-flipping hard-label attacks can be understood as approximating the true gradient sign. Guided by this principled analysis, we propose a novel attack framework featuring a zero-query initialization strategy and a Pattern-Driven Optimization (PDO)
algorithm. We provide theoretical guarantees that our initialization yields higher cosine similarity to the true gradient sign than random baselines, and our PDO module achieves significantly lower query complexity than baseline search methods. Extensive experiments across CIFAR-10, ImageNet, and ObjectNet-covering standard and adversarially trained models, commercial APIs, and CLIP models-demonstrate that our method consistently outperforms SOTA hard-label attacks in both success rate and efficiency, particularly under low query budgets. Furthermore, our method demonstrates robust generalization across corrupted data (ImageNet-C), biomedical images (PathMNIST), and dense prediction tasks such as
segmentation. Notably, it bypasses the stateful defense Blacklight, achieving a 0% detection rate.
Read the full abstract ▾
All authors Jun Liu, Leo Yu Zhang, Fengpeng Li, Isao Echizen, Jiantao Zhou
Machine Learning
Federated Learning for the Design of Parametric Insurance Indices under Heterogeneous Renewable Production Losses
Fallou Niakh
We propose a
federated learning framework for the calibration of parametric insurance indices under heterogeneous renewable energy production losses. Producers locally model their losses using Tweedie generalized linear models and private data, while a common index is learned through
federated optimization without sharing raw observations. The approach accommodates heterogeneity in variance and link functions and directly minimizes a global deviance objective in a distributed setting. We establish theoretical guarantees under sub-exponential
covariate distributions, showing that the Lipschitz constants of the local Tweedie objectives scale as $\frac{1}{\phi_i}$, where $\phi_i$ is the dispersion parameter of producer $i$. This heterogeneity in smoothness causes naive
federated averaging to be biased toward producers with stable microclimates — precisely those least in need of basis-risk protection — and motivates the use of corrected aggregation schemes. We implement and compare FedAvg, FedProx and FedOpt, and
benchmark them against an existing approximation-based aggregation method. A progressive pool expansion experiment involving up to 121
solar farms in Germany reveals that the approximation-based method becomes entirely non-computable beyond the pool of 50 farms, while
federated learning remains valid and actively improves as the pool grows.
Federated learning is also over $250\times$ faster than the approximation-based
benchmark, establishing it as the only computationally and statistically valid approach for heterogeneous producer pools.
Read the full abstract ▾
Machine Learning · Cross-listed from math.NA
Streaming Operator Inference for Model Reduction of Large-Scale Dynamical Systems
Tomoki Koike, Prakash Mohan, Marc T. Henry de Frahan et al. (+2 more)
Projection-based model reduction enables efficient simulation of complex dynamical systems by constructing low-dimensional surrogate models from high-dimensional data. The Operator Inference (OpInf) approach learns such reduced surrogate models through a two-step process: constructing a low-dimensional basis via Singular Value Decomposition (SVD) to compress the data, then solving a linear least-squares (LS) problem to infer reduced operators that govern the dynamics in this compressed space, all without access to the underlying code or full model operators, i.e., non-intrusively. Traditional OpInf operates as a batch learning method, where both the SVD and LS steps process all data simultaneously. This poses a barrier to deployment of the approach on large-scale applications where
dataset sizes prevent the loading of all data into
memory at once. Additionally, the traditional batch approach does not naturally allow model updates using new data acquired during online computation. To address these limitations, we propose Streaming OpInf, which learns reduced models from sequentially arriving data streams. Our approach employs incremental SVD for adaptive basis construction and recursive LS for streaming operator updates, eliminating the need to store complete data sets while enabling online model adaptation. The approach can flexibly combine different choices of streaming
algorithms for numerical linear
algebra: we systematically explore the impact of these choices both analytically and numerically to identify effective combinations for accurate reduced model learning. Numerical experiments on
benchmark problems and a large-scale turbulent channel flow demonstrate that Streaming OpInf achieves accuracy comparable to batch OpInf while reducing
memory requirements by over 99% and enabling dimension reductions exceeding 31,000x, resulting in orders-of-magnitude faster predictions.
Read the full abstract ▾
All authors Tomoki Koike, Prakash Mohan, Marc T. Henry de Frahan, Julie Bessac, Elizabeth Qian
Machine Learning
Threshold Differential Attention for Sink-Free, Ultra-Sparse, and Non-Dispersive Language Modeling
Xingyue Huang, Xueying Ding, Mingxuan Ju et al. (+3 more)
Softmax
attention struggles with long contexts due to structural limitations: the strict sum-to-one constraint forces
attention sinks on irrelevant
tokens, and probability mass disperses as sequence lengths increase. We tackle these problems with Threshold Differential
Attention (TDA), a sink-free
attention mechanism that achieves ultra-sparsity and improved robustness at longer sequence lengths without the computational overhead of projection methods or the performance degradation caused by noise accumulation of standard rectified
attention. TDA applies row-wise extreme-value thresholding with a length-dependent gate, retaining only exceedances. Inspired by the differential
transformer, TDA also subtracts an inhibitory view to enhance expressivity. Theoretically, we prove that TDA controls the expected number of spurious survivors per row to $O(1)$ and that consensus spurious matches across independent views vanish as context grows. Empirically, TDA produces $>99\%$ exact zeros and eliminates
attention sinks while maintaining competitive performance on standard and long-context
benchmarks.
Read the full abstract ▾
All authors Xingyue Huang, Xueying Ding, Mingxuan Ju, Yozen Liu, Neil Shah, Tong Zhao
Machine Learning
EMoE: Eigenbasis-Guided Routing for Mixture-of-Experts
Anzhe Cheng, Shukai Duan, Shixuan Li et al. (+5 more)
The relentless scaling of deep learning models has led to unsustainable computational demands, positioning Mixture-of-Experts (MoE) architectures as a promising path towards greater efficiency. However, MoE models are plagued by two fundamental challenges: 1) a load imbalance problem known as the“rich get richer" phenomenon, where a few experts are over-utilized, and 2) an expert homogeneity problem, where experts learn redundant representations, negating their purpose. Current solutions typically employ an auxiliary load-balancing loss that, while mitigating imbalance, often exacerbates homogeneity by enforcing uniform routing at the expense of specialization. To resolve this, we introduce the Eigen-Mixture-of-Experts (EMoE), a novel architecture that leverages a routing mechanism based on a learned orthonormal eigenbasis. EMoE projects input
tokens onto this shared eigenbasis and routes them based on their
alignment with the principal components of the feature space. This principled,
geometric partitioning of data intrinsically promotes both balanced expert utilization and the development of diverse, specialized experts, all without the need for a conflicting auxiliary loss function. Our code is publicly available at https://github.com/Belis0811/EMoE.
Read the full abstract ▾
All authors Anzhe Cheng, Shukai Duan, Shixuan Li, Chenzhong Yin, Mingxi Cheng, Shahin Nazarian, Paul Thompson, Paul Bogdan
Machine Learning
SolarGPT-QA: A Domain-Adaptive Large Language Model for Educational Question Answering in Space Weather and Heliophysics
Santosh Chapagain, MohammadReza EskandariNasab, Onur Vural et al. (+2 more)
Solar activity, including
solar flares, coronal mass ejections (CMEs), and geomagnetic storms can significantly impact satellites, aviation, power grids, data centers, and space missions. Extreme
solar events can cause substantial economic damage with limited advance warning, underscoring the importance of early warning systems, accurate
forecasting, and effective education in space science. Although large
language models (LLMs) perform well on general tasks, they often lack domain specific knowledge and pedagogical capability to clearly explain complex space science concepts. We introduce SolarGPT-QA, a question answering system based on a domain adapted large
language model built on the LLaMA-3 base model. The model is trained using scientific literature and large scale question and answer data generated with GPT-4 and refined using Grok-3 in a student friendly storytelling style. To evaluate response quality, we employ an LLM-as-judge evaluation framework, where a strong reference model assesses generated answers using structured criteria including scientific accuracy, clarity, completeness, and pedagogical effectiveness. Results show that SolarGPT-QA performs strongly relative to general purpose models in zero shot settings and achieves competitive performance compared to instruction tuned models for educational explanations in space weather and heliophysics. Ablation studies indicate that combining domain adaptive pretraining with fine tuning is important for balancing scientific accuracy and educational effectiveness.
Read the full abstract ▾
All authors Santosh Chapagain, MohammadReza EskandariNasab, Onur Vural, Shah Muhammad Hamdi, Soukaina Filali Boubrahimi
Machine Learning
SynQP: A Framework and Metrics for Evaluating the Quality and Privacy Risk of Synthetic Data
Bing Hu, Yixin Li, Asma Bahamyirou et al. (+1 more)
The use of synthetic data in health applications raises
privacy concerns, yet the lack of open frameworks for
privacy evaluations has slowed its adoption. A major challenge is the absence of accessible
benchmark datasets for evaluating
privacy risks, due to difficulties in acquiring sensitive data. To address this, we introduce SynQP, an open framework for benchmarking
privacy in synthetic data generation (SDG) using simulated sensitive data, ensuring that original data remains confidential. We also highlight the need for
privacy metrics that fairly account for the probabilistic nature of machine learning models. As a demonstration, we use SynQP to
benchmark CTGAN and propose a new identity disclosure risk metric that offers a more accurate estimation of
privacy risks compared to existing approaches. Our work provides a critical tool for improving the transparency and reliability of
privacy evaluations, enabling safer use of synthetic data in health-related applications. % In our quality evaluations, non-private models achieved near-perfect machine-learning efficacy \(\ge0.97\). Our
privacy assessments (Table II) reveal that DP consistently lowers both identity disclosure risk (SD-IDR) and membership-inference attack risk (SD-MIA), with all DP-augmented models staying below the 0.09 regulatory threshold. Code available at https://github.com/CAN-SYNH/SynQP
Read the full abstract ▾
All authors Bing Hu, Yixin Li, Asma Bahamyirou, Helen Chen
Machine Learning · Cross-listed from math.OC
Offline Policy Learning with Weight Clipping and Heaviside Composite Optimization
Jingren Liu, Hanzhang Qin, Junyi Liu et al. (+2 more)
Offline policy learning aims to use historical data to learn an optimal personalized decision rule. In the standard estimate-then-optimize framework, reweighting-based methods (e.g., inverse propensity weighting or doubly robust
estimators) are widely used to produce unbiased estimates of policy values. However, when the propensity scores of some treatments are small, these reweighting-based methods suffer from high variance in policy value estimation, which may mislead the downstream policy optimization and yield a learned policy with inferior value. In this paper, we systematically develop an offline policy learning
algorithm based on a weight-clipping
estimator that truncates small propensity scores via a clipping threshold chosen to minimize the mean squared error (MSE) in policy value estimation. Focusing on linear policies, we address the bilevel and discontinuous objective induced by weight-clipping-based policy optimization by reformulating the problem as a Heaviside composite optimization problem, which provides a rigorous computational framework. The reformulated policy optimization problem is then solved efficiently using the progressive integer programming method, making practical policy learning tractable. We establish an upper bound for the suboptimality of the proposed
algorithm, which reveals how the reduction in MSE of policy value estimation, enabled by our proposed weight-clipping
estimator, leads to improved policy learning performance.
Read the full abstract ▾
All authors Jingren Liu, Hanzhang Qin, Junyi Liu, Mabel C. Chou, Jong-Shi Pang
Machine Learning
Neural Isomorphic Fields: A Transformer-based Algebraic Numerical Embedding
Hamidreza Sadeghi, Saeedeh Momtazi and Reza Safabakhsh
Neural network models often face challenges when processing very small or very large numbers due to issues such as overflow, underflow, and unstable output variations. To mitigate these problems, we propose using
embedding vectors for numbers instead of directly using their raw values. These
embeddings aim to retain essential algebraic properties while preventing numerical instabilities. In this paper, we introduce, for the first time, a fixed-length number
embedding vector that preserves algebraic operations, including addition, multiplication, and comparison, within the field of rational numbers. We propose a novel Neural Isomorphic Field, a neural abstraction of algebraic structures such as groups and fields. The elements of this neural field are
embedding vectors that maintain algebraic structure during computations. Our experiments demonstrate that addition performs exceptionally well, achieving over 95 percent accuracy on key algebraic tests such as identity, closure, and associativity. In contrast, multiplication exhibits challenges, with accuracy ranging from 53 percent to 73 percent across various algebraic properties. These findings highlight the model's strengths in preserving algebraic properties under addition while identifying avenues for further improvement in handling multiplication.
Read the full abstract ▾
Machine Learning
PTL-PINNs: Perturbation-Guided Transfer Learning with Physics- Informed Neural Networks for Nonlinear Systems
Duarte Alexandrino, Ben Moseley and Pavlos Protopapas
Accurately and efficiently solving nonlinear differential equations is crucial for modeling dynamic behavior across science and engineering. Physics-Informed Neural Networks (PINNs) have emerged as a powerful solution that embeds physical laws in training by enforcing equation residuals. However, these struggle to model nonlinear dynamics, suffering from limited generalization across problems and long training times. To address these limitations, we propose a perturbation-guided transfer learning framework for PINNs (PTL-PINN), which integrates perturbation theory with transfer learning to efficiently solve nonlinear equations. Unlike gradient-based transfer learning, PTL-PINNs solve an approximate linear perturbative system using closed-form expressions, enabling rapid generalization with the time complexity of
matrix-vector multiplication. We show that PTL-PINNs achieve accuracy comparable to various Runge-Kutta methods, with computational speeds up to one order of magnitude faster. To
benchmark performance, we solve a broad set of problems, including nonlinear oscillators across various damping regimes, the
equilibrium-centered Lotka-Volterra system, the KPP-Fisher and the Wave equation. Since perturbation theory sets the accuracy bound of PTL-PINNs, we systematically evaluate its practical applicability. This work connects long-standing perturbation methods with PINNs, demonstrating how perturbation theory can guide foundational models to solve nonlinear systems with speeds comparable to those of classical solvers.
Read the full abstract ▾
Machine Learning
Mitigating Cultural Bias in LLMs via Multi-Agent Cultural Debate
Qian Tan, Lei Jiang, Yuting Zeng et al. (+2 more)
Large
language models (LLMs) exhibit systematic Western-centric bias, yet whether prompting in non-Western
languages (e.g., Chinese) can mitigate this remains understudied. Answering this question requires rigorous evaluation and effective mitigation, but existing approaches fall short on both fronts: evaluation methods force outputs into predefined cultural categories without a neutral option, while mitigation relies on expensive multi-cultural corpora or
agent frameworks that use functional roles (e.g., Planner–Critique) lacking explicit cultural representation. To address these gaps, we introduce CEBiasBench, a Chinese–English bilingual
benchmark, and Multi-
Agent Vote (MAV), which enables explicit “no bias” judgments. Using this framework, we find that Chinese prompting merely shifts bias toward East Asian perspectives rather than eliminating it. To mitigate such persistent bias, we propose Multi-
Agent Cultural Debate (MACD), a training-free framework that assigns
agents distinct cultural personas and orchestrates deliberation via a "Seeking Common Ground while Reserving Differences" strategy. Experiments demonstrate that MACD achieves 57.6% average No Bias Rate evaluated by LLM-as-judge and 86.0% evaluated by MAV (vs. 47.6% and 69.0% baseline using GPT-4o as backbone) on CEBiasBench and generalizes to the Arabic CAMeL
benchmark, confirming that explicit cultural representation in
agent frameworks is essential for cross-cultural
fairness.
Read the full abstract ▾
All authors Qian Tan, Lei Jiang, Yuting Zeng, Shuoyang Ding, Xiaohua Xu
Machine Learning
Learning to Factorize and Adapt: A Versatile Approach Toward Universal Spatio-Temporal Foundation Models
Siru Zhong, Junjie Qiu, Yangyu Wu et al. (+7 more)
Spatio-Temporal (ST) Foundation Models (STFMs) promise cross-
dataset generalization, yet joint ST pretraining is computationally expensive and grapples with the heterogeneity of domain-specific spatial patterns. Substantially extending our preliminary conference version, we present FactoST-v2, an enhanced factorized framework redesigned for full weight transfer and arbitrary-length generalization. FactoST-v2 decouples universal temporal learning from domain-specific spatial adaptation. The first stage pretrains a minimalist encoder-only backbone using randomized sequence masking to capture invariant temporal dynamics, enabling probabilistic
quantile prediction across variable horizons. The second stage employs a streamlined adapter to rapidly inject spatial awareness via meta adaptive learning and prompting. Comprehensive evaluations across diverse domains demonstrate that FactoST-v2 achieves state-of-the-art accuracy with linear efficiency - significantly outperforming existing foundation models in zero-shot and few-shot scenarios while rivaling domain-specific expert baselines. This factorized paradigm offers a practical, scalable path toward truly universal STFMs. Code is available at https://github.com/CityMind-Lab/FactoST.
Read the full abstract ▾
All authors Siru Zhong, Junjie Qiu, Yangyu Wu, Yiqiu Liu, Yuanpeng He, Zhongwen Rao, Bin Yang, Chenjuan Guo, Hao Xu, Yuxuan Liang
Machine Learning · Cross-listed from cs.CV
Automating Parameter Selection in Deep Image Prior for Fluorescence Microscopy Image Denoising via Similarity-Based Parameter Transfer
Lina Meyer, Felix Wissel, Tobias Knopp et al. (+8 more)
Unsupervised deep image prior (DIP) addresses shortcomings of training data requirements and limited generalization associated with supervised deep learning. The performance of DIP depends on the network architecture and the stopping point of its iterative process. Optimizing these parameters for a new image requires time, restricting DIP application in domains where many images need to be processed. Focusing on fluorescence microscopy data, we hypothesize that similar images share comparable optimal parameter configurations for DIP-based denoising, potentially enabling optimization-free DIP for fluorescence microscopy. We generated a calibration (n=110) and validation set (n=55) of semantically different images from an open-source
dataset for a network architecture search targeted towards ideal U-net architectures and stopping points. The calibration set represented our transfer basis. The validation set enabled the assessment of which image similarity criterion yields the best results. We then implemented AUTO-DIP, a pipeline for automatic parameter transfer, and compared it to the originally published DIP configuration (baseline) and a state-of-the-art image-specific variational denoising approach. We show that a parameter transfer from the calibration
dataset to a test image based on only image metadata similarity (e.g., microscope type, imaged specimen) leads to similar and better performance than a transfer based on quantitative image similarity measures. AUTO-DIP outperforms the baseline DIP (DIP with original DIP parameters) as well as the variational denoising approaches for several open-source test
datasets of varying complexity, particularly for very noisy inputs. Applications to locally acquired fluorescence microscopy images further proved superiority of AUTO-DIP.
Read the full abstract ▾
All authors Lina Meyer, Felix Wissel, Tobias Knopp, Susanne Pfefferle, Ralf Fliegert, Maximilian Sandmann, Liana Uebler, Franziska Möckl, Björn-Philipp Diercks, David Lohr, René Werner
Machine Learning · Cross-listed from q-bio.NC
A New Strategy for Artificial Intelligence: Training Foundation Models Directly on Human Brain Data
Maël Donoso
While foundation models have achieved remarkable results across a diversity of domains, they still rely on human-generated data, such as text, as a fundamental source of knowledge. However, this data is ultimately the product of human
brains, the filtered projection of a deeper neural complexity. In this paper, we explore a new strategy for artificial intelligence: moving beyond surface-level statistical regularities by training foundation models directly on human
brain data. We hypothesize that neuroimaging data could open a window into elements of human cognition that are not accessible through observable actions, and argue that this additional knowledge could be used, alongside classical training data, to overcome some of the current limitations of foundation models. While previous research has demonstrated the possibility to train classical machine learning, deep learning, or reinforcement learning models on neural patterns, this path remains largely unexplored for high-level cognitive functions. Here, we classify the current limitations of foundation models, as well as the promising
brain regions and cognitive processes that could be leveraged to address them, along four levels: perception, valuation, execution, and integration. Then, we propose two general methods that could be implemented to prioritize the use of limited neuroimaging data for strategically chosen, high-value steps in foundation model training: reinforcement learning from human
brain (RLHB) and chain of thought from human
brain (CoTHB). We also discuss the potential implications for
agents, artificial general intelligence, and artificial superintelligence, as well as the ethical, social, and technical challenges and opportunities.
Read the full abstract ▾
Machine Learning · Cross-listed from econ.EM
Nonlinear Dynamic Factor Analysis With a Transformer Network
Oliver Snellman
The paper develops a
Transformer architecture for estimating dynamic factors from multivariate time series data under flexible identification assumptions. Performance on small
datasets is improved substantially by using a conventional factor model as prior information via a regularization term in the training objective. The results are interpreted with
Attention matrices that quantify the relative importance of variables and their lags for the factor estimate. Time variation in
Attention patterns can help detect regime switches and evaluate narratives. Monte Carlo experiments suggest that the
Transformer is more accurate than the linear factor model, when the data deviate from linear-Gaussian assumptions. An empirical application uses the
Transformer to construct a coincident index of U.S. real economic activity.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.NE
Speaking to Silicon: Neural Communication with Bitcoin Mining ASICs
Francisco Angulo de Lafuente, Vladimir Veselov and Richard Goodman
This definitive research memoria presents a comprehensive, mathematically verified paradigm for neural communication with Bitcoin mining Application-Specific Integrated
Circuits (ASICs), integrating five complementary frameworks: thermodynamic reservoir computing, hierarchical number system theory, algorithmic analysis, network latency optimization, and machine-checked mathematical formalization. We establish that obsolete cryptocurrency mining hardware exhibits emergent computational properties enabling bidirectional information exchange between AI systems and silicon substrates. The research program demonstrates: (1) reservoir computing with NARMA-10 Normalized Root Mean Square Error (NRMSE) of 0.8661; (2) the Thermodynamic Probability Filter (TPF) achieving 92.19% theoretical energy reduction; (3) the Virtual Block Manager achieving +25% effective hashrate; and (4) hardware universality across multiple ASIC families including Antminer S9, Lucky Miner LV06, and Goldshell LB-Box. A significant contribution is the machine-checked mathematical formalization using Lean 4 and Mathlib, providing unambiguous definitions, machine-verified
theorems, and reviewer-proof claims. Key
theorems proven include: independence implies zero leakage, predictor beats baseline implies non-independence (the logical core of TPF), energy savings theoretical maximum, and Physical Unclonable Function (PUF) distinguishability witnesses. Vladimir Veselov's hierarchical number system theory explains why early-round information contains predictive power. This work establishes a new paradigm: treating ASICs not as passive computational substrates but as active conversational partners whose thermodynamic state encodes exploitable computational information.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ML
A Kernel Approach for Semi-implicit Variational Inference
Longlin Yu, Ziheng Cheng, Shiyue Zhang et al. (+1 more)
Semi-implicit variational inference (SIVI) enhances the expressiveness of variational families through hierarchical semi-implicit distributions, but the intractability of their densities makes standard ELBO-based optimization biased. Recent score-matching approaches to SIVI (SIVI-SM) address this issue via a minimax formulation, at the expense of an additional lower-level optimization problem. In this paper, we propose kernel semi-implicit variational inference (KSIVI), a principled and tractable alternative that eliminates the lower-level optimization by leveraging kernel methods. We show that when optimizing over a reproducing kernel
Hilbert space, the lower-level problem admits an explicit solution, reducing the objective to the kernel Stein discrepancy (KSD). Exploiting the hierarchical structure of semi-implicit distributions, the resulting KSD objective can be efficiently optimized using
stochastic gradient methods. We establish optimization guarantees via variance bounds on Monte Carlo gradient
estimators and derive statistical generalization bounds of order $\tilde{\mathcal{O}}(1/\sqrt{n})$. We further introduce a multi-layer hierarchical extension that improves expressiveness while preserving tractability. Empirical results on synthetic and real-world
Bayesian inference tasks demonstrate the effectiveness of KSIVI.
Read the full abstract ▾
All authors Longlin Yu, Ziheng Cheng, Shiyue Zhang, Cheng Zhang
Machine Learning
Why Loss Re-weighting Works If You Stop Early: Training Dynamics of Unconstrained Features
Yize Zhao and Christos Thrampoulidis
The application of loss reweighting in modern deep learning presents a nuanced picture. While it fails to alter the terminal learning phase in overparameterized deep neural networks (DNNs) trained on high-dimensional
datasets, empirical evidence consistently shows it offers significant benefits early in training. To transparently demonstrate and analyze this phenomenon, we introduce a small-scale model (SSM). This model is specifically designed to abstract the inherent complexities of both the DNN architecture and the input data, while maintaining key information about the structure of imbalance within its spectral components. On the one hand, the SSM reveals how vanilla empirical risk minimization preferentially learns to distinguish majority classes over minorities early in training, consequently delaying minority learning. In stark contrast, reweighting restores balanced learning dynamics, enabling the simultaneous learning of features associated with both majorities and minorities.
Read the full abstract ▾
Machine Learning
Extreme Value Policy Optimization for Safe Reinforcement Learning
Shiqing Gao, Yihang Zhou, Shuai Shao et al. (+5 more)
Ensuring safety is a critical challenge in applying Reinforcement Learning (RL) to real-world scenarios. Constrained Reinforcement Learning (CRL) addresses this by maximizing returns under predefined constraints, typically formulated as the expected cumulative cost. However, expectation-based constraints overlook rare but high-impact extreme value events in the tail distribution, such as black swan incidents, which can lead to severe constraint violations. To address this issue, we propose the Extreme Value policy Optimization (EVO)
algorithm, leveraging Extreme Value Theory (EVT) to model and exploit extreme reward and cost samples, reducing constraint violations. EVO introduces an extreme
quantile optimization objective to explicitly capture extreme samples in the cost tail distribution. Additionally, we propose an extreme prioritization mechanism during replay, amplifying the learning signal from rare but high-impact extreme samples. Theoretically, we establish upper bounds on expected constraint violations during policy updates, guaranteeing strict constraint satisfaction at a zero-violation
quantile level. Further, we demonstrate that EVO achieves a lower probability of constraint violations than expectation-based methods and exhibits lower variance than
quantile regression methods. Extensive experiments show that EVO significantly reduces constraint violations during training while maintaining competitive policy performance compared to baselines.
Read the full abstract ▾
All authors Shiqing Gao, Yihang Zhou, Shuai Shao, Haoyu Luo, Yiheng Bing, Jiaxin Ding, Luoyi Fu, Xinbing Wang
Machine Learning · Cross-listed from cs.AI
Kernel-Based Learning of Safety Barriers
Oliver Schön, Zhengang Zhong and Sadegh Soudjani
The rapid integration of AI
algorithms in safety-critical applications such as autonomous driving and healthcare is raising significant concerns about the ability to meet stringent safety standards. Traditional tools for formal safety verification struggle with the black-box nature of AI-driven systems and lack the flexibility needed to scale to the complexity of real-world applications. In this paper, we present a data-driven approach for safety verification and synthesis of black-box systems with discrete-time
stochastic dynamics. We employ the concept of control barrier certificates, which can guarantee safety of the system, and learn the certificate directly from a set of system trajectories. We use conditional mean
embeddings to embed data from the system into a reproducing kernel
Hilbert space (RKHS) and construct an RKHS ambiguity set that can be inflated to robustify the result to out-of-distribution behavior. We provide the theoretical results on how to apply the approach to general classes of temporal logic specifications beyond safety. For the data-driven computation of safety barriers, we leverage a finite
Fourier expansion to cast a typically intractable semi-infinite optimization problem as a linear program. The resulting spectral barrier allows us to leverage the fast
Fourier transform to generate the relaxed problem efficiently, offering a scalable yet distributionally robust framework for verifying safety. Our work moves beyond restrictive assumptions on system dynamics and uncertainty, as demonstrated on two case studies including a black-box system with a neural network controller.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CR
MongoDB Injection Query Classification Model using MongoDB Log files as Training Data
Shaunak Perni, Minal Shirodkar and Ramdas Karmalli
NoSQL Injection attacks are a class of cybersecurity attacks where an attacker sends a specifically engineered query to a NoSQL database which then performs an unauthorized operation. To defend against such attacks, rule based systems were initially developed but then were found to be ineffective to innovative injection attacks hence a model based approach was developed. Most model based detection systems, during testing gave exponentially positive results but were trained only on the query statement sent to the server. However due to the scarcity of data and class imbalances these model based systems were found to be not effective against all attacks in the real world. This paper explores classifying NoSQL injection attacks sent to a MongoDB server based on Log Data, and other extracted features excluding raw query statements. The log data was collected from a simulated attack on an empty MongoDB server which was then processed and explored. A discriminant analysis was carried out to determine statistically significant features to discriminate between injection and benign queries resulting in a
dataset of significant features. Several Machine learning based classification models using an AutoML library, "FLAML", as well as 6 manually programmed models were trained on this
dataset , which were then trained on 50 randomized samples of data, cross validated and evaluated. The study found that the best model was the "FLAML" library's "XGBoost limited depth" model with an accuracy of 71%.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.MM
Learning Audio-Visual Embeddings with Inferred Latent Interaction Graphs
Donghuo Zeng, Hao Niu, Yanan Wang et al. (+1 more)
Learning robust audio-visual
embeddings requires bringing genuinely related audio and visual signals together while filtering out incidental co-occurrences - background noise, unrelated elements, or unannotated events. Most contrastive and triplet-loss methods use sparse annotated labels per clip and treat any co-occurrence as semantic similarity. For example, a
video labeled "train" might also contain motorcycle audio and visual, because "motorcycle" is not the chosen annotation; standard methods treat these co-occurrences as negatives to true motorcycle anchors elsewhere, creating false negatives and missing true cross-modal dependencies. We propose a framework that leverages soft-label predictions and inferred latent interactions to address these issues: (1) Audio-Visual Semantic
Alignment Loss (AV-SAL) trains a teacher network to produce aligned soft-label distributions across modalities, assigning nonzero probability to co-occurring but unannotated events and enriching the supervision signal. (2) Inferred Latent Interaction Graph (ILI) applies the GRaSP
algorithm to teacher soft labels to infer a sparse, directed dependency graph among classes. This graph highlights directional dependencies (e.g., "Train (visual)" -> "Motorcycle (audio)") that expose likely semantic or conditional relationships between classes; these are interpreted as estimated dependency patterns. (3) Latent Interaction Regularizer (LIR): A student network is trained with both metric loss and a regularizer guided by the ILI graph, pulling together
embeddings of dependency-linked but unlabeled pairs in proportion to their soft-label probabilities. Experiments on AVE and VEGAS
benchmarks show consistent improvements in mean average precision (mAP), demonstrating that integrating inferred latent interactions into
embedding learning enhances robustness and semantic coherence.
Read the full abstract ▾
All authors Donghuo Zeng, Hao Niu, Yanan Wang, Masato Taya
Machine Learning · Cross-listed from cs.AI
Process In-Context Learning: Enhancing Mathematical Reasoning via Dynamic Demonstration Insertion
Ang Gao, Changshuo Zhang, Xiao Zhang et al. (+4 more)
In-context learning (ICL) has proven highly effective across diverse large
language model (LLM) tasks. However, its potential for enhancing tasks that demand step-by-step logical deduction, such as mathematical
reasoning, remains underexplored. A core limitation of existing ICL approaches is their static use of demonstrations: examples are pre-selected before inference and remain fixed, failing to adapt to the dynamic confusion points that often arise during multi-step
reasoning such as ambiguous calculations or logical gaps. These unresolved confusion points can lead to cascading errors that degrade final accuracy. To tackle this issue, we propose Process In-Context Learning (PICL), a dynamic demonstration integration framework designed to boost mathematical
reasoning by responding to real-time inference needs. PICL operates in two stages: 1)~it identifies potential confusion points by analyzing semantics and
entropy in the
reasoning process and summarizes their core characteristics; 2)~upon encountering these points, it retrieves relevant demonstrations from the demonstration pool that match the confusion context and inserts them directly into the ongoing
reasoning process to guide subsequent steps. Experiments show that PICL outperforms baseline methods by mitigating mid-inference confusion, highlighting the value of adaptive demonstration insertion in complex mathematical
reasoning.
Read the full abstract ▾
All authors Ang Gao, Changshuo Zhang, Xiao Zhang, Deyang Li, Minjun Zhao, Fangchao Liu, Xinyu Zhang
Machine Learning
One-Shot Price Forecasting with Covariate-Guided Experts under Privacy Constraints
Ren He, Yinliang Xu, Jinfeng Wang et al. (+2 more)
Forecasting in power systems often involves multivariate time series with complex dependencies and strict
privacy constraints across regions. Traditional
forecasting methods require significant expert knowledge and struggle to generalize across diverse deployment scenarios. Recent advancements in pre-trained time series models offer new opportunities, but their zero-shot performance on domain-specific tasks remains limited. To address these challenges, we propose a novel MoE Encoder module that augments pretrained
forecasting models by injecting a sparse mixture-of-experts layer between tokenization and encoding. This design enables two key capabilities: (1) trans forming multivariate
forecasting into an expert-guided univariate task, allowing the model to effectively capture inter-variable relations, and (2) supporting localized training and lightweight parameter sharing in
federated settings where raw data cannot be exchanged. Extensive experiments on public multivariate
datasets demonstrate that MoE-Encoder significantly improves
forecasting accuracy compared to strong baselines. We further simulate
federated environments and show that transferring only MoE-Encoder parameters allows efficient adaptation to new regions, with minimal performance degradation. Our findings suggest that MoE-Encoder provides a scalable and
privacy-aware extension to foundation time series models.
Read the full abstract ▾
All authors Ren He, Yinliang Xu, Jinfeng Wang, Jeremy Watson, Jian Song
Machine Learning
R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning
Jingchu Wang, Bingbing Xu, Yige Yuan et al. (+4 more)
Existing reinforcement learning methods for LLM
reasoning implicitly assume that the policy generating training trajectories should coincide with the one producing inference responses. We argue that this is a misleading inductive bias: the optimization-optimal trajectory distribution favors informative gradients, whereas the inference-optimal response distribution emphasizes accuracy and consistency. Forcing both into a single policy entangles their gradients and suppresses exploration. We propose R$^2$PO (Residual Rollout Policy Optimization), which attaches a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, diversifying rollouts during training while keeping inference generation intact. Experiments show that R$^2$PO consistently outperforms baselines, with average accuracy gains of 3.4% on MATH-500 and 1.3% on APPS, alongside more diverse rollouts and reduced length bias. Our code is available at https://github.com/RRPO-ARR/Code.
Read the full abstract ▾
All authors Jingchu Wang, Bingbing Xu, Yige Yuan, Dan Zhang, Bin Xie, Xiaoqian Sun, Huawei Shen