Peer Revenue Equity in science

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Machine Learning

cs.LG Watch Machine Learning (sign in to use)

Friday 16 January

Machine Learning · Cross-listed from math.NA

Exact Constraint Enforcement in Physics-Informed Extreme Learning Machines using Null-Space Projection Framework

Physics-informed extreme learning machines (PIELMs) typically impose boundary and initial conditions through penalty terms, yielding only approximate satisfaction that is sensitive to user-specified weights and can propagate errors into the interior solution. This work introduces Null-Space Projected PIELM (NP-PIELM), achieving exact constraint enforcement through algebraic projection in coefficient space. The method exploits the geometric structure of the admissible coefficient manifold, recognizing that it admits a decomposition through the null space of the boundary operator. By characterizing this manifold via a translation-invariant representation and projecting onto the kernel component, optimization is restricted to constraint-preserving directions, transforming the constrained problem into unconstrained least-squares where boundary conditions are satisfied exactly at discrete collocation points. This eliminates penalty coefficients, dual variables, and problem-specific constructions while preserving single-shot training efficiency. Numerical experiments on elliptic and parabolic problems including complex geometries and mixed boundary conditions validate the framework.
Read the full abstract

All authors Rishi Mishra, Smriti, Balaji Srinivasan, Sundararajan Natarajan, Ganapathy Krishnamurthi

Machine Learning · Cross-listed from stat.ML

Memorize Early, Then Query: Inlier-Memorization-Guided Active Outlier Detection

Outlier detection (OD) aims to identify abnormal instances, known as outliers or anomalies, by learning typical patterns of normal data, or inliers. Performing OD under an unsupervised regime-without any information about anomalous instances in the training data-is challenging. A recently observed phenomenon, known as the inlier-memorization (IM) effect, where deep generative models (DGMs) tend to memorize inlier patterns during early training, provides a promising signal for distinguishing outliers. However, existing unsupervised approaches that rely solely on the IM effect still struggle when inliers and outliers are not well-separated or when outliers form dense clusters. To address these limitations, we incorporate active learning to selectively acquire informative labels, and propose IMBoost, a novel framework that explicitly reinforces the IM effect to improve outlier detection. Our method consists of two stages: 1) a warm-up phase that induces and promotes the IM effect, and 2) a polarization phase in which actively queried samples are used to maximize the discrepancy between inlier and outlier scores. In particular, we propose a novel query strategy and tailored loss function in the polarization phase to effectively identify informative samples and fully leverage the limited labeling budget. We provide a theoretical analysis showing that the IMBoost consistently decreases inlier risk while increasing outlier risk throughout training, thereby amplifying their separation. Extensive experiments on diverse benchmark datasets demonstrate that IMBoost not only significantly outperforms state-of-the-art active OD methods but also requires substantially less computational cost.
Read the full abstract

Machine Learning

Constant Metric Scaling in Riemannian Computation

Constant rescaling of a Riemannian metric appears in many computational settings, often through a global scale parameter that is introduced either explicitly or implicitly. Although this operation is elementary, its consequences are not always made clear in practice and may be confused with changes in curvature, manifold structure, or coordinate representation. In this note we provide a short, self-contained account of constant metric scaling on arbitrary Riemannian manifolds. We distinguish between quantities that change under such a scaling, including norms, distances, volume elements, and gradient magnitudes, and geometric objects that remain invariant, such as the Levi–Civita connection, geodesics, exponential and logarithmic maps, and parallel transport. We also discuss implications for Riemannian optimization, where constant metric scaling can often be interpreted as a global rescaling of step sizes rather than a modification of the underlying geometry. The goal of this note is purely expository and is intended to clarify how a global metric scale parameter can be introduced in Riemannian computation without altering the geometric structures on which these methods rely.
Read the full abstract

Machine Learning

Reasoning Distillation for Lightweight Automated Program Repair

We study whether lightweight symbolic reasoning supervision can improve fix type classification in compact automated program repair models. Small code models are attractive for resource-constrained settings, but they typically produce only a single prediction, making it unclear whether they learn meaningful program structure or rely on shallow correlations. We propose a reasoning distillation approach in which a large teacher model provides structured symbolic reasoning tags alongside fix-type labels. These tags capture high-level causal properties of bugs without relying on free-form explanations. We train a CodeT5-based student model under label-only and reasoning-distilled settings on the IntroClass benchmark. Reasoning supervision consistently improves macro averaged performance, particularly on less frequent bug categories, without increasing model size or complexity. We further analyze the relationship between reasoning accuracy and fix-type prediction, showing that correct reasoning traces strongly correlate with correct predictions, while not fully determining them. Our results suggest that symbolic reasoning distillation is a practical way to improve interpretability and robustness in lightweight program repair models.
Read the full abstract

Machine Learning

Toward Adaptive Grid Resilience: A Gradient-Free Meta-RL Framework for Critical Load Restoration

Restoring critical loads after extreme events demands adaptive control to maintain distribution-grid resilience, yet uncertainty in renewable generation, limited dispatchable resources, and nonlinear dynamics make effective restoration difficult. Reinforcement learning (RL) can optimize sequential decisions under uncertainty, but standard RL often generalizes poorly and requires extensive retraining for new outage configurations or generation patterns. We propose a meta-guided gradient-free RL (MGF-RL) framework that learns a transferable initialization from historical outage experiences and rapidly adapts to unseen scenarios with minimal task-specific tuning. MGF-RL couples first-order meta-learning with evolutionary strategies, enabling scalable policy search without gradient computation while accommodating nonlinear, constrained distribution-system dynamics. Experiments on IEEE 13-bus and IEEE 123-bus test systems show that MGF-RL outperforms standard RL, MAML-based meta-RL, and model predictive control across reliability, restoration speed, and adaptation efficiency under renewable forecast errors. MGF-RL generalizes to unseen outages and renewable patterns while requiring substantially fewer fine-tuning episodes than conventional RL. We also provide sublinear regret bounds that relate adaptation efficiency to task similarity and environmental variation, supporting the empirical gains and motivating MGF-RL for real-time load restoration in renewable-rich distribution grids.
Read the full abstract

Machine Learning

Noise-Driven Exploration and Transient Freezing Select Flat Minima in Stochastic Gradient Descent

Stochastic gradient descent (SGD) is central to deep learning, yet the dynamical origin of its preference for flatter, more generalizable solutions remains unclear. Here, by analyzing SGD learning dynamics, we identify a nonequilibrium mechanism that governs solution selection during training. Numerical experiments reveal a transient exploratory phase in which SGD trajectories repeatedly escape sharp valleys and migrate toward flatter regions of the loss landscape before becoming confined to a final basin. Using a tractable physical model, we show that SGD noise reshapes the loss landscape into an effective potential that preferentially stabilizes flat solutions. We further uncover a transient freezing mechanism: as training progresses, the flattening landscape suppresses transitions between competing valleys. Stronger SGD noise delays this freezing transition, prolonging the exploratory phase and thereby increasing the probability of convergence to flatter minima. Together, these results provide a unified physical framework connecting learning dynamics, loss-landscape geometry, and generalization, and suggest guiding principles for the design of more effective optimization algorithms.
Read the full abstract

All authors Ning Yang, Yikuan Zhang, Qi Ouyang, Chao Tang, Yuhai Tu

Machine Learning

Multivariate LSTM-Based Forecasting for Renewable Energy: Enhancing Climate Change Mitigation

The increasing integration of renewable energy sources (RESs) into modern power systems presents significant opportunities but also notable challenges, primarily due to the inherent variability of RES generation. Accurate forecasting of RES generation is crucial for maintaining the reliability, stability, and economic efficiency of power system operations. Traditional approaches, such as deterministic methods and stochastic programming, frequently depend on representative scenarios generated through clustering techniques like K-means. However, these methods may fail to fully capture the complex temporal dependencies and non-linear patterns within RES data. This paper introduces a multivariate Long Short-Term Memory (LSTM)-based network designed to forecast RESs generation using their real-world historical data. The proposed model effectively captures long-term dependencies and interactions between different RESs, utilizing historical data from both local and neighboring areas to enhance predictive accuracy. In the case study, we showed that the proposed forecasting approach results in lower CO2 emissions, and a more reliable supply of electric loads.
Read the full abstract

Machine Learning · Cross-listed from q-bio.QM

Depression Detection Based on Electroencephalography Using a Hybrid Deep Neural Network CNN-GRU and MRMR Feature Selection

This study investigates the detection and classification of depressive and non-depressive states using deep learning approaches. Depression is a prevalent mental health disorder that substantially affects quality of life, and early diagnosis can greatly enhance treatment effectiveness and patient care. However, conventional diagnostic methods rely heavily on self-reported assessments, which are often subjective and may lack reliability. Consequently, there is a strong need for objective and accurate techniques to identify depressive states. In this work, a deep learning based framework is proposed for the early detection of depression using EEG signals. EEG data, which capture underlying brain activity and are not influenced by external behavioral factors, can reveal subtle neural changes associated with depression. The proposed approach combines convolutional neural networks (CNNs) and gated recurrent units (GRUs) to jointly extract spatial and temporal features from EEG recordings. The minimum redundancy maximum relevance (MRMR) algorithm is then applied to select the most informative features, followed by classification using a fully connected neural network. The results demonstrate that the proposed model achieves high performance in accurately identifying depressive states, with an overall accuracy of 98.74%. By effectively integrating temporal and spatial information and employing optimized feature selection, this method shows strong potential as a reliable tool for clinical applications. Overall, the proposed framework not only enables accurate early detection of depression but also has the potential to support improved treatment strategies and patient outcomes.
Read the full abstract

Machine Learning

HOSL: Hybrid-Order Split Learning for Memory-Constrained Edge Training

Split learning (SL) enables collaborative training of large language models (LLMs) between resource-constrained edge devices and compute-rich servers by partitioning model computation across the network boundary. However, existing SL systems predominantly rely on first-order (FO) optimization, which requires clients to store intermediate quantities such as activations for backpropagation. This results in substantial memory overhead, largely negating benefits of model partitioning. In contrast, zeroth-order (ZO) optimization eliminates backpropagation and significantly reduces memory usage, but often suffers from slow convergence and degraded performance. In this work, we propose HOSL, a novel Hybrid-Order Split Learning framework that addresses this fundamental trade-off between memory efficiency and optimization effectiveness by strategically integrating ZO optimization on the client side with FO optimization on the server side. By employing memory-efficient ZO gradient estimation at the client, HOSL eliminates backpropagation and activation storage, reducing client memory consumption. Meanwhile, server-side FO optimization ensures fast convergence and competitive performance. Theoretically, we show that HOSL achieves an $\mathcal{O}(\sqrt{d_c/TQ})$ rate, which depends on client-side model dimension $d_c$ rather than the full model dimension $d$, demonstrating that convergence improves as more computation is offloaded to the server. Extensive experiments on OPT models (125M and 1.3B parameters) across 6 tasks demonstrate that HOSL reduces client GPU memory by up to 3.7$\times$ compared to the FO method while achieving accuracy within 0.20%-4.23% of this baseline. Furthermore, HOSL outperforms the ZO baseline by up to 15.55%, validating the effectiveness of our hybrid strategy for memory-efficient training on edge devices.
Read the full abstract

All authors Aakriti Lnu, Zhe Li, Dandan Liang, Chao Huang, Rui Li, Haibo Yang

Machine Learning · Cross-listed from cs.CV

RobuMTL: Enhancing Multi-Task Learning Robustness Against Weather Conditions

Robust Multi-Task Learning (MTL) is crucial for autonomous systems operating in real-world environments, where adverse weather conditions can severely degrade model performance and reliability. In this paper, we introduce RobuMTL, a novel architecture designed to adaptively address visual degradation by dynamically selecting task-specific hierarchical Low-Rank Adaptation (LoRA) modules and a LoRA expert squad based on input perturbations in a mixture-of-experts fashion. Our framework enables adaptive specialization based on input characteristics, improving robustness across diverse real-world conditions. To validate our approach, we evaluated it on the PASCAL and NYUD-v2 datasets and compared it against single-task models, standard MTL baselines, and state-of-the-art methods. On the PASCAL benchmark, RobuMTL delivers a +2.8% average relative improvement under single perturbations and up to +44.4% under mixed weather conditions compared to the MTL baseline. On NYUD-v2, RobuMTL achieves a +9.7% average relative improvement across tasks. The code is available at GitHub.
Read the full abstract

Machine Learning · Cross-listed from cs.CV

Self-learned representation-guided latent diffusion model for breast cancer classification in deep ultraviolet whole surface images

Breast-Conserving Surgery (BCS) requires precise intraoperative margin assessment to preserve healthy tissue. Deep Ultraviolet Fluorescence Scanning Microscopy (DUV-FSM) offers rapid, high-resolution surface imaging for this purpose; however, the scarcity of annotated DUV data hinders the training of robust deep learning models. To address this, we propose an Self-Supervised Learning (SSL)-guided Latent Diffusion Model (LDM) to generate high-quality synthetic training patches. By guiding the LDM with embeddings from a fine-tuned DINO teacher, we inject rich semantic details of cellular structures into the synthetic data. We combine real and synthetic patches to fine-tune a Vision Transformer (ViT), utilizing patch prediction aggregation for WSI-level classification. Experiments using 5-fold cross-validation demonstrate that our method achieves 96.47 % accuracy and reduces the FID score to 45.72, significantly outperforming class-conditioned baselines.
Read the full abstract

All authors Pouya Afshin, David Helminiak, Tianling Niu, Julie M. Jorns, Tina Yen, Bing Yu, Dong Hye Ye

Machine Learning · Cross-listed from cs.IT

A PAC-Bayesian Analysis of Channel-Induced Degradation in Edge Inference

In the emerging paradigm of edge learning, neural networks (NNs) are partitioned across distributed edge devices that collaboratively perform inference via wireless transmission. However, deploying NNs for edge inference over wireless channels inevitably leads to performance degradation, as the exact channel realizations in the inference stage are not known in the training stage. In this paper, we establish a theoretical framework to evaluate and bound this performance degradation. Inspired by statistical learning theory, we define a wireless generalization error to characterize the gap between the empirical performance during training and the expected inference performance under the true stochastic channel. To enable theoretical analysis, we introduce an augmented NN model that incorporates channel statistics directly into the weight space. Leveraging the PAC-Bayesian framework, we derive a high-probability bound on this error, which provides theoretical guarantees for wireless inference performance. Furthermore, we propose a channel-aware training algorithm that minimizes a tractable surrogate objective based on the derived bound. Simulations demonstrate that the proposed algorithm effectively improves wireless inference performance and model robustness under various channel conditions.
Read the full abstract

Machine Learning

FAConvLSTM: Factorized-Attention ConvLSTM for Efficient Feature Extraction in Multivariate Climate Data

Learning physically meaningful spatiotemporal representations from high-resolution multivariate Earth observation data is challenging due to strong local dynamics, long-range teleconnections, multi-scale interactions, and nonstationarity. While ConvLSTM2D is a commonly used baseline, its dense convolutional gating incurs high computational cost and its strictly local receptive fields limit the modeling of long-range spatial structure and disentangled climate dynamics. To address these limitations, we propose FAConvLSTM, a Factorized-Attention ConvLSTM layer designed as a drop-in replacement for ConvLSTM2D that simultaneously improves efficiency, spatial expressiveness, and physical interpretability. FAConvLSTM factorizes recurrent gate computations using lightweight [1 times 1] bottlenecks and shared depthwise spatial mixing, substantially reducing channel complexity while preserving recurrent dynamics. Multi-scale dilated depthwise branches and squeeze-and-excitation recalibration enable efficient modeling of interacting physical processes across spatial scales, while peephole connections enhance temporal precision. To capture teleconnection-scale dependencies without incurring global attention cost, FAConvLSTM incorporates a lightweight axial spatial attention mechanism applied sparsely in time. A dedicated subspace head further produces compact per timestep embeddings refined through temporal self-attention with fixed seasonal positional encoding. Experiments on multivariate spatiotemporal climate data shows superiority demonstrating that FAConvLSTM yields more stable, interpretable, and robust latent representations than standard ConvLSTM, while significantly reducing computational overhead.
Read the full abstract

Thursday 15 January

Machine Learning · Cross-listed from cs.NI

Large Language Model (LLM)-enabled Reinforcement Learning for Wireless Network Optimization

Enhancing future wireless networks presents a significant challenge for networking systems due to diverse user demands and the emergence of 6G technology. While reinforcement learning (RL) is a powerful framework, it often encounters difficulties with high-dimensional state spaces and complex environments, leading to substantial computational demands, distributed intelligence, and potentially inconsistent outcomes. Large language models (LLMs), with their extensive pretrained knowledge and advanced reasoning capabilities, offer promising tools to enhance RL in optimizing 6G wireless networks. We explore RL models augmented by LLMs, emphasizing their roles and the potential benefits of their synergy in wireless network optimization. We then examine LLM-enabled RL across various protocol layers: physical, data link, network, transport, and application layers. Additionally, we propose an LLM-assisted state representation and semantic extraction to enhance the multi-agent reinforcement learning (MARL) framework. This approach is applied to service migration and request routing, as well as topology graph generation in unmanned aerial vehicle (UAV)-satellite networks. Through case studies, we demonstrate that our framework effectively performs optimization of wireless network. Finally, we outline prospective research directions for LLM-enabled RL in wireless network optimization.
Read the full abstract

All authors Jie Zheng, Ruichen Zhang, Dusit Niyato, Haijun Zhang, Jiacheng Wang, Hongyang Du, Jiawen Kang, Zehui Xiong

Machine Learning · Cross-listed from q-fin.CP

Deep g-Pricing for CSI 300 Index Options with Volatility Trajectories and Market Sentiment

Option pricing in real markets faces fundamental challenges. The Black–Scholes–Merton (BSM) model assumes constant volatility and uses a linear generator $g(t,x,y,z)=-ry$, while lacking explicit behavioral factors, resulting in systematic departures from observed dynamics. This paper extends the BSM model by learning a nonlinear generator within a deep Forward–Backward Stochastic Differential Equation (FBSDE) framework. We propose a dual-network architecture where the value network $u_\theta$ learns option prices and the generator network $g_\phi$ characterizes the pricing mechanism, with the hedging strategy $Z_t=\sigma_t X_t \nabla_x u_\theta$ obtained via automatic differentiation. The framework adopts forward recursion from a learnable initial condition $Y_0=u_\theta(0,\cdot)$, naturally accommodating volatility trajectory and sentiment features. Empirical results on CSI 300 index options show that our method reduces Mean Absolute Error (MAE) by 32.2% and Mean Absolute Percentage Error (MAPE) by 35.3% compared with BSM. Interpretability analysis indicates that architectural improvements are effective across all option types, while the information advantage is asymmetric between calls and puts. Specifically, call option improvements are primarily driven by sentiment features, whereas put options show more balanced contributions from volatility trajectory and sentiment features. This finding aligns with economic intuition regarding option pricing mechanisms.
Read the full abstract

All authors Yilun Zhang, Zheng Tang, Hexiang Sun, Yufeng Shi

Machine Learning

Latent Structural Similarity Networks for Unsupervised Discovery in Multivariate Time Series

This paper proposes a task-agnostic discovery layer for multivariate time series that constructs a relational hypothesis graph over entities without assuming linearity, stationarity, or a downstream objective. The method learns window-level sequence representations using an unsupervised sequence-to-sequence autoencoder, aggregates these representations into entity-level embeddings, and induces a sparse similarity network by thresholding a latent-space similarity measure. This network is intended as an analyzable abstraction that compresses the pairwise search space and exposes candidate relationships for further investigation, rather than as a model optimized for prediction, trading, or any decision rule. The framework is demonstrated on a challenging real-world dataset of hourly cryptocurrency returns, illustrating how latent similarity induces coherent network structure; a classical econometric relation is also reported as an external diagnostic lens to contextualize discovered edges.
Read the full abstract

Machine Learning

Activation Sensitivity as a Unifying Principle for Post-Training Quantization

Post-training quantization (PTQ) methods for large language models compress pretrained weights to low bit-widths without retraining by optimizing layer-local proxy objectives. Activation-aware scaling methods such as AWQ protect input channels with large activation norms, whereas second-order error-compensation methods such as GPTQ redistribute rounding error using the inverse of the unweighted input covariance matrix. We show that both paradigms – along with diagonal Fisher quantization, block reconstruction, and Hadamard rotation methods – are special cases of minimizing a single loss-derived quadratic form governed by gradient-weighted activation sensitivity. From a Gauss-Newton expansion of the task loss at a linear layer $Y = X W^\top$, we prove that under isotropic output curvature the expected excess loss from weight perturbation $\Delta W$ reduces to $\frac{1}{2 d_{\mathrm{out}}}\operatorname{Tr}(\Delta W H_D \Delta W^\top)$, where $H_D = \frac{1}{n} X^\top D X$ weights each calibration token's outer product by its squared output-gradient norm $D_{tt} = \|\nabla_{y_t}\mathcal{L}\|_2^2$. Within this framework, AWQ corresponds to a diagonal approximation of $H_D$ with uniform token weights ($D = I_n$) optimized via channel scaling, while GPTQ corresponds to Cholesky error compensation on $H_D$ under $D = I_n$. We derive closed-form optimal scaling exponents for diagonal sensitivity protection, formulate sensitivity-weighted Cholesky quantization (SAQ-GPTQ) on $H_D$, and prove bounds on token-gradient heterogeneity bias, cross-layer error amplification across residual blocks, and finite-sample calibration concentration. Controlled numerical simulations confirm that under token-gradient heterogeneity, replacing $X^\top X$ with $X^\top D X$ preserves $>99\%$ top-channel saliency recall and reduces INT3/INT4 excess loss by roughly $3\times$ over unweighted GPTQ.
Read the full abstract

Machine Learning

Machine learning model for predicting surface wettability in laser-textured metal alloys

Surface wettability, governed by both topography and chemistry, plays a critical role in applications such as heat transfer, lubrication, microfluidics, and surface coatings. In this study, we present a machine learning (ML) framework capable of accurately predicting the wettability of laser-textured metal alloys using experimentally derived morphological and chemical features. Superhydrophilic and superhydrophobic surfaces were fabricated on AA6061 and AISI 4130 alloys via nanosecond laser texturing followed by chemical immersion treatments. Surface morphology was quantified using the Laws texture energy method and profilometry, while surface chemistry was characterized through X-ray photoelectron spectroscopy (XPS), extracting features such as functional group polarity, molecular volume, and peak area fraction. These features were used to train an ensemble neural network model incorporating residual connections, batch normalization, and dropout regularization. The model achieved high predictive accuracy (R2 = 0.942, RMSE = 13.896), outperforming previous approaches. Feature importance analysis revealed that surface chemistry had the strongest influence on contact angle prediction, with topographical features also contributing significantly. This work demonstrates the potential of artificial intelligence to model and predict wetting behavior by capturing the complex interplay of surface characteristics, offering a data-driven pathway for designing tailored functional surfaces.
Read the full abstract

All authors Mohammad Mohammadzadeh Sanandaji, Danial Ebrahimzadeh, Mohammad Ikram Haider, Yaser Mike Banad, Aleksandar Poleksic, Hongtao Ding

Machine Learning · Cross-listed from cs.CV

Zeros can be Informative: Masked Binary U-Net for Image Segmentation on Tensor Cores

Real-time image segmentation is a key enabler for AR/VR, robotics, drones, and autonomous systems, where tight accuracy, latency, and energy budgets must be met on resource-constrained edge devices. While U-Net offers a favorable balance of accuracy and efficiency compared to large transformer-based models, achieving real-time performance on high-resolution input remains challenging due to compute, memory, and power limits. Extreme quantization, particularly binary networks, is appealing for its hardware-friendly operations. However, two obstacles limit practicality: (1) severe accuracy degradation, and (2) a lack of end-to-end implementations that deliver efficiency on general-purpose GPUs. We make two empirical observations that guide our design. (1) An explicit zero state is essential: training with zero masking to binary U-Net weights yields noticeable sparsity. (2) Quantization sensitivity is uniform across layers. Motivated by these findings, we introduce Masked Binary U-Net (MBU-Net), obtained through a cost-aware masking strategy that prioritizes masking where it yields the highest accuracy-per-cost, reconciling accuracy with near-binary efficiency. To realize these gains in practice, we develop a GPU execution framework that maps MBU-Net to Tensor Cores via a subtractive bit-encoding scheme, efficiently implementing masked binary weights with binary activations. This design leverages native binary Tensor Core BMMA instructions, enabling high throughput and energy savings on widely available GPUs. Across 3 segmentation benchmarks, MBU-Net attains near full-precision accuracy (3% average drop) while delivering 2.04x speedup and 3.54x energy reductions over a 16-bit floating point U-Net.
Read the full abstract

All authors Chunshu Wu, Ruibing Song, Sushant Kondguli, Tong Geng, Ang Li

Machine Learning · Cross-listed from cs.SE

The Llama 4 Herd: Architecture, Training, Evaluation, and Deployment Notes

This document consolidates publicly reported technical details about Metas Llama 4 model family. It summarizes (i) released variants (Scout and Maverick) and the broader herd context including the previewed Behemoth teacher model, (ii) architectural characteristics beyond a high-level MoE description covering routed/shared-expert structure, early-fusion multimodality, and long-context design elements reported for Scout (iRoPE and length generalization strategies), (iii) training disclosures spanning pre-training, mid-training for long-context extension, and post-training methodology (lightweight SFT, online RL, and lightweight DPO) as described in release materials, (iv) developer-reported benchmark results for both base and instruction-tuned checkpoints, and (v) practical deployment constraints observed across major serving environments, including provider-specific context limits and quantization packaging. The manuscript also summarizes licensing obligations relevant to redistribution and derivative naming, and reviews publicly described safeguards and evaluation practices. The goal is to provide a compact technical reference for researchers and practitioners who need precise, source-backed facts about Llama 4.
Read the full abstract

Machine Learning

Size is Not the Solution: Deformable Convolutions for Effective Physics Aware Deep Learning

Physics-aware deep learning (PADL) enables rapid prediction of complex physical systems, yet current convolutional neural network (CNN) architectures struggle with highly nonlinear flows. While scaling model size addresses complexity in broader AI, this approach yields diminishing returns for physics modeling. Drawing inspiration from Hybrid Lagrangian-Eulerian (HLE) numerical methods, we introduce deformable physics-aware recurrent convolutions (D-PARC) to overcome the rigidity of CNNs. Across Burgers' equation, Navier-Stokes, and reactive flows, D-PARC achieves superior fidelity compared to substantially larger architectures. Analysis reveals that kernels display anti-clustering behavior, evolving into a learned "active filtration" strategy distinct from traditional h- or p-adaptivity. Effective receptive field analysis confirms that D-PARC autonomously concentrates resources in high-strain regions while coarsening focus elsewhere, mirroring adaptive refinement in computational mechanics. This demonstrates that physically intuitive architectural design can outperform parameter scaling, establishing that strategic learning in lean networks offers a more effective path forward for PADL than indiscriminate network expansion.
Read the full abstract

All authors Jack T. Beerman, Shobhan Roy, H.S. Udaykumar, Stephen S. Baek

Machine Learning · Cross-listed from q-bio.NC

AI Agents Need Memory Control Over More Context

AI agents are increasingly used in long, multi-turn workflows in both research and enterprise settings. As interactions grow, agent behavior often degrades due to loss of constraint focus, error accumulation, and memory-induced drift. This problem is especially visible in real-world deployments where context evolves, distractions are introduced, and decisions must remain consistent over time. A common practice is to equip agents with persistent memory through transcript replay or retrieval-based mechanisms. While convenient, these approaches introduce unbounded context growth and are vulnerable to noisy recall and memory poisoning, leading to unstable behavior and increased drift. In this work, we introduce the Agent Cognitive Compressor (ACC), a bio-inspired memory controller that replaces transcript replay with a bounded internal state updated online at each turn. ACC separates artifact recall from state commitment, enabling stable conditioning while preventing unverified content from becoming persistent memory. We evaluate ACC using an agent-judge-driven live evaluation framework that measures both task outcomes and memory-driven anomalies across extended interactions. Across scenarios spanning IT operations, cybersecurity response, and healthcare workflows, ACC consistently maintains bounded memory and exhibits more stable multi-turn behavior, with significantly lower hallucination and drift than transcript replay and retrieval-based agents. These results show that cognitive compression provides a practical and effective foundation for reliable memory control in long-horizon AI agents.
Read the full abstract

Machine Learning · Cross-listed from cs.SE

Reinforcement Learning for Dynamic Workflow Optimization in CI/CD Pipelines

Continuous Integration and Continuous Deployment (CI/CD) pipelines are central to modern software delivery, yet their static workflows often introduce inefficiencies as systems scale. This paper proposes a reinforcement learning (RL) based approach to dynamically optimize CI/CD pipeline workflows. The pipeline is modeled as a Markov Decision Process, and an RL agent is trained to make runtime decisions such as selecting full, partial, or no test execution in order to maximize throughput while minimizing testing overhead. A configurable CI/CD simulation environment is developed to evaluate the approach across build, test, and deploy stages. Experimental results show that the RL optimized pipeline achieves up to a 30 percent improvement in throughput and approximately a 25 percent reduction in test execution time compared to static baselines, while maintaining a defect miss rate below 5 percent. The agent learns to selectively skip or abbreviate tests for low risk commits, accelerating feedback cycles without significantly increasing failure risk. These results demonstrate the potential of reinforcement learning to enable adaptive and intelligent DevOps workflows, providing a practical pathway toward more efficient, resilient, and sustainable CI/CD automation.
Read the full abstract

All authors Aniket Abhishek Soni, Milan Parikh, Rashi Nimesh Kumar Dhenia, Jubin Abhishek Soni, Ayush Raj Jha, Sneja Mitinbhai Shah

Machine Learning

Realistic Curriculum Reinforcement Learning for Autonomous and Sustainable Marine Vessel Navigation

Sustainability is becoming increasingly critical in the maritime transport, encompassing both environmental and social impacts, such as Greenhouse Gas (GHG) emissions and navigational safety. Traditional vessel navigation heavily relies on human experience, often lacking autonomy and emission awareness, and is prone to human errors that may compromise safety. In this paper, we propose a Curriculum Reinforcement Learning (CRL) framework integrated with a realistic, data-driven marine simulation environment and a machine learning-based fuel consumption prediction module. The simulation environment is constructed using real-world vessel movement data and enhanced with a Diffusion Model to simulate dynamic maritime conditions. Vessel fuel consumption is estimated using historical operational data and learning-based regression. The surrounding environment is represented as image-based inputs to capture spatial complexity. We design a lightweight, policy-based CRL agent with a comprehensive reward mechanism that considers safety, emissions, timeliness, and goal completion. This framework effectively handles complex tasks progressively while ensuring stable and efficient learning in continuous action spaces. We validate the proposed approach in a sea area of the Indian Ocean, demonstrating its efficacy in enabling sustainable and safe vessel navigation.
Read the full abstract

All authors Zhang Xiaocai, Xiao Zhe, Liang Maohan, Liu Tao, Li Haijiang, Zhang Wenbin

Machine Learning

Action Shapley: A Training Data Selection Metric for World Model in Reinforcement Learning

Numerous offline and model-based reinforcement learning systems incorporate world models to emulate the inherent environments. A world model is particularly important in scenarios where direct interactions with the real environment is costly, dangerous, or impractical. The efficacy and interpretability of such world models are notably contingent upon the quality of the underlying training data. In this context, we introduce Action Shapley as an agnostic metric for the judicious and unbiased selection of training data. To facilitate the computation of Action Shapley, we present a randomized dynamic algorithm specifically designed to mitigate the exponential complexity inherent in traditional Shapley value computations. Through empirical validation across five data-constrained real-world case studies, the algorithm demonstrates a computational efficiency improvement exceeding 80% in comparison to conventional exponential time computations. Furthermore, our Action Shapley-based training data selection policy consistently outperforms ad-hoc training data selection.
Read the full abstract

Machine Learning · Cross-listed from physics.plasm-ph

Learning collision operators from plasma phase space data using differentiable simulators

We propose a methodology to infer collision operators from phase space data of plasma dynamics. Our approach combines a differentiable kinetic simulator, whose core component in this work is a differentiable Fokker-Planck solver, with a gradient-based optimisation method to learn the collisional operators that best describe the phase space dynamics. We test our method using data from two-dimensional Particle-in-Cell simulations of spatially uniform thermal plasmas, and learn the collision operator that captures the self-consistent electromagnetic interaction between finite-size charged particles over a wide variety of simulation parameters. We demonstrate that the learned operators are more accurate than alternative estimates based on particle tracks, while making no prior assumptions about the relevant time scales of the processes and significantly reducing memory requirements. We find that the retrieved operators, obtained in the non-relativistic regime, are in excellent agreement with theoretical predictions derived for electrostatic scenarios. Our results show that differentiable simulators offer a powerful and computational efficient approach to infer novel operators for a wide rage of problems, such as electromagnetically dominated collisional dynamics and stochastic wave-particle interactions.
Read the full abstract

All authors Diogo D. Carvalho, Pablo J. Bilbao, Warren B. Mori, Luis O. Silva, E. Paulo Alves

Machine Learning

Unit-Consistent (UC) Adjoint for GSD and Backprop in Deep Learning Applications

Deep neural networks constructed from linear maps and positively homogeneous nonlinearities (e.g., ReLU) possess a fundamental gauge symmetry: the network function is invariant to node-wise diagonal rescalings. However, standard gradient descent is not equivariant to this symmetry, causing optimization trajectories to depend heavily on arbitrary parameterizations. Prior work has proposed rescaling-invariant optimization schemes for positively homogeneous networks (e.g., path-based or path-space updates). Our contribution is complementary: we formulate the invariance requirement at the level of the backward adjoint/optimization geometry, which provides a simple, operator-level recipe that can be applied uniformly across network components and optimizer state. By replacing the Euclidean transpose with a Unit-Consistent (UC) adjoint, we derive UC gauge-consistent steepest descent and backprogation.
Read the full abstract

Machine Learning

Beyond Accuracy: A Stability-Aware Metric for Multi-Horizon Forecasting

Traditional time series forecasting methods optimize for accuracy alone. This objective neglects temporal consistency, in other words, how consistently a model predicts the same future event as the forecast origin changes. We introduce the forecast accuracy and coherence score (forecast AC score for short) for measuring the quality of probabilistic multi-horizon forecasts in a way that accounts for both multi-horizon accuracy and stability. Our score additionally allows user-specified weights to balance accuracy and consistency requirements. As an example application, we implement the score as a differentiable objective function for training seasonal auto-regressive integrated models and evaluate it on the M4 Hourly benchmark dataset. Results demonstrate consistent improvements over traditional maximum likelihood estimation. Regarding stability, the AC-optimized model generated out-of-sample forecasts with 15.8% reduced variance over forecasts targeting the same timestamp. In terms of accuracy, the AC-optimized model achieved considerable improvements for medium-to-long-horizon forecasts. While one-step-ahead forecasts exhibited a 3.9% increase in MSE, forecasts from horizon three onward experienced improved accuracy, with a peak improvement of approximately 6% in MSE at horizons 9-12. These results indicate that our metric successfully trains models to produce more stable and accurate multi-step forecasts in exchange for a relatively small degradation in one-step-ahead performance.
Read the full abstract

All authors Chutian Ma, Grigorii Pomazkin, Giacinto Paolo Saggese, Paul Smith

Machine Learning

AI-Guided Human-In-the-Loop Inverse Design of High Performance Engineering Structures

Inverse design tools such as Topology Optimization (TO) can achieve new levels of improvement for high-performance engineered structures. However, widespread use is hindered by high computational times and a black-box nature that inhibits user interaction. Human-in-the-loop TO approaches are emerging that integrate human intuition into the design generation process. However, these rely on the time-consuming bottleneck of iterative region selection for design modifications. To reduce the number of iterative trials, this contribution presents an AI co-pilot that uses machine learning to predict the user's preferred regions. The prediction model is configured as an image segmentation task with a U-Net architecture. It is trained on synthetic datasets where human preferences either identify the longest topological member or the most complex structural connection. The model successfully predicts plausible regions for modification and presents them to the user as AI recommendations. The human preference model demonstrates generalization across diverse and non-standard TO problems and exhibits emergent behavior outside the single-region selection training data. Demonstration examples show that the new human-in-the-loop TO approach that integrates the AI co-pilot can improve manufacturability or improve the linear buckling load by 39% while only increasing the total design time by 15 sec compared to conventional simplistic TO.
Read the full abstract

All authors Dat Quoc Ha, Md Ferdous Alam, Markus J. Buehler, Faez Ahmed, Josephine V. Carstensen

Machine Learning · Cross-listed from cs.CL

Reasoning Models Generate Societies of Thought

Large language models have achieved remarkable capabilities across domains, yet mechanisms underlying sophisticated reasoning remain elusive. Recent reasoning models outperform comparable instruction-tuned models on complex cognitive tasks, attributed to extended computation through longer chains of thought. Here we show that enhanced reasoning emerges not from extended computation alone, but from simulating multi-agent-like interactions – a society of thought – which enables diversification and debate among internal cognitive perspectives characterized by distinct personality traits and domain expertise. Through quantitative analysis and mechanistic interpretability methods applied to reasoning traces, we find that reasoning models like DeepSeek-R1 and QwQ-32B exhibit much greater perspective diversity than instruction-tuned models, activating broader conflict between heterogeneous personality- and expertise-related features during reasoning. This multi-agent structure manifests in conversational behaviors, including question-answering, perspective shifts, and the reconciliation of conflicting views, and in socio-emotional roles that characterize sharp back-and-forth conversations, together accounting for the accuracy advantage in reasoning tasks. Controlled reinforcement learning experiments reveal that base models increase conversational behaviors when rewarded solely for reasoning accuracy, and fine-tuning models with conversational scaffolding accelerates reasoning improvement over base models. These findings indicate that the social organization of thought enables effective exploration of solution spaces. We suggest that reasoning models establish a computational parallel to collective intelligence in human groups, where diversity enables superior problem-solving when systematically structured, which suggests new opportunities for agent organization to harness the wisdom of crowds.
Read the full abstract

All authors Junsol Kim, Shiyang Lai, Nino Scherrer, Blaise Agüera y Arcas, James Evans