# Research Papers — 2026-01 · bundle page 79 of 189 > Full-text markdown of papers #3901–3950 of 9403 published in 2026-01, newest first. AI-readable mirrors of the full 9403-paper 2026-01 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-01/page/79 [← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/78.md) [Next bundle →](https://aipod.dev/papers/archive/2026-01/page/80.md) ### Test-Time Adaptation for Non-stationary Time Series: From Synthetic Regime Shifts to Financial Markets Authors: Yurui Wu, Qingying Deng, Wonou Chung, Mairui Li Published: 2026-01-20 Categories: q-fin.ST, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2602.00073v1 Time series encountered in practice are rarely stationary. When the data distribution changes, a forecasting model trained on past observations can lose accuracy. We study a small-footprint test-time adaptation (TTA) framework for causal timeseries forecasting and direction classification. The backbone is frozen, and only normalization affine parameters are updated using recent unlabeled windows. For classification we minimize entropy and enforce temporal consistency; for regression we minimize prediction variance across weak time-preserving augmentations and optionally distill from an EMA teacher. A quadratic drift penalty and an uncertainty triggered fallback keep updates stable. We evaluate this framework in two stages: synthetic regime shifts on ETT benchmarks, and daily equity and FX series (SPY, QQQ, EUR/USD) across pandemic, high-inflation, and recovery regimes. On synthetic gradual drift, normalization-based TTA improves forecasting error, while in financial markets a simple batch-normalization statistics update is a robust default and more aggressive norm-only adaptation can even hurt. Our results provide practical guidance for deploying TTA on non-stationary time series. --- ### Generative AI-enhanced Probabilistic Multi-Fidelity Surrogate Modeling Via Transfer Learning Authors: Jice Zeng, David Barajas-Solano, Hui Chen Published: 2026-01-20 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2602.00072v2 The performance of machine learning surrogates is critically dependent on data quality and quantity. This presents a major challenge, as high-fidelity (HF) data is often scarce and computationally expensive to acquire, while low-fidelity (LF) data is abundant but less accurate. To address this data-scarcity problem, we propose a probabilistic multi-fidelity surrogate modeling framework that integrates transfer learning with generative modeling. We employ a normalizing flow (NF) generative model as the backbone, which is trained in two phases: (i) the NF is first pretrained on a large LF dataset to learn a probabilistic forward model; (ii) the pretrained model is then fine-tuned on a small HF dataset, allowing it to correct for LF--HF discrepancies via knowledge transfer. To relax the dimension-preserving constraint of standard bijective NFs, we integrate surjective (dimension-reducing) layers with standard coupling blocks. This architecture enables learned dimension reduction while preserving the ability to train with exact likelihoods. The resulting surrogate provides fast probabilistic predictions with quantified uncertainty and significantly outperforms LF-only baselines while using fewer HF evaluations. We validate the approach for two benchmark systems: a rail-sleeper-ballast and a reinforced concrete slab. For both applications, we combine many coarse-mesh (LF) simulations with a limited set of fine-mesh (HF) simulations. The proposed model achieves probabilistic predictions with HF accuracy, demonstrating a practical path toward data-efficient, generative AI-driven surrogates for complex engineering systems. --- ### FoundationalASSIST: An Educational Dataset for Foundational Knowledge Tracing and Pedagogical Grounding of LLMs Authors: Eamon Worden, Cristina Heffernan, Neil Heffernan, Shashank Sonkar Published: 2026-01-20 Categories: cs.CY, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2602.00070v1 Can Large Language Models understand how students learn? As LLMs are deployed for adaptive testing and personalized tutoring, this question becomes urgent -- yet we cannot answer it with existing resources. Current educational datasets provide only question identifiers and binary correctness labels, rendering them opaque to LLMs that reason in natural language. We address this gap with FoundationalASSIST, the first English educational dataset providing the complete information needed for research on LLMs in education: full question text, actual student responses (not just right/wrong), records of which wrong answers students chose, and alignment to Common Core K-12 standards. These 1.7 million interactions from 5,000 students enable research directions that were previously impossible to pursue, from fine-tuning student models to analyzing misconception patterns. To demonstrate the dataset's utility, we evaluate four frontier models (GPT-OSS-120B, Llama-3.3-70B, Qwen3-Next-80B variants) on two complementary task families: Knowledge Tracing, testing whether LLMs can predict student performance on questions, and the exact answer a student will give; and \textbf{Pedagogical Grounding}, testing whether LLMs understand the properties that make assessment items effective. Our evaluation reveals significant gaps in current LLM capabilities. Every model barely achieves a trivial baseline on knowledge tracing. All models fall below random chance on item discrimination, indicating that LLMs do not understand what makes one problem more diagnostic than another. Models do show competence at judging relative difficulty (up to 68.6%), but this partial success only highlights the gaps elsewhere. These results establish that substantial advances are needed before LLMs can reliably support personalized learning at scale. We release FoundationalASSIST to support progress on these foundational challenges. --- ### On finite-dimensional encoding/decoding theorems for neural operators Authors: Vinícius Luz Oliveira, Vladimir G. Pestov Published: 2026-01-20 Categories: math.FA, cs.LG arXiv: https://arxiv.org/abs/2602.00068v1 Recently, versions of neural networks with infinite-dimensional affine operators inside the computational units (``neural operator'' networks) have been applied to learn solutions to differential equations. To enable practical computations, one employs finite-dimensional encoding/decoding theorems of the following kind: every continuous mapping $f$ between function spaces $E$ and $F$ is approximated in the topology of uniform convergence on compacta by continuous mappings factoring through two finite dimensional Banach spaces. Such a result is known (Kovachki et al., 2023) for $E,F$ being Banach spaces having the approximation property. We point out that the result needs no assumptions on $E,F$ whatsoever and remains true not only for all normed spaces, but for arbitrary locally convex spaces as well. At the same time, an analogous result for $C^k$-smooth mappings and the $C^k$ compact open topology, $k\geq 1$, holds if and only if the space $E$ has the approximation property. This analysis may be useful already because non-normable locally convex function spaces are common in the theory of differential equations, the main field of applications for the emerging theory. --- ### Modality as Heterogeneity: Node Splitting and Graph Rewiring for Multimodal Graph Learning Authors: Yihan Zhang, Ercan E. Kuruoglu Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.00067v1 Multimodal graphs are gaining increasing attention due to their rich representational power and wide applicability, yet they introduce substantial challenges arising from severe modality confusion. To address this issue, we propose NSG (Node Splitting Graph)-MoE, a multimodal graph learning framework that integrates a node-splitting and graph-rewiring mechanism with a structured Mixture-of-Experts (MoE) architecture. It explicitly decomposes each node into modality-specific components and assigns relation-aware experts to process heterogeneous message flows, thereby preserving structural information and multimodal semantics while mitigating the undesirable mixing effects commonly observed in general-purpose GNNs. Extensive experiments on three multimodal benchmarks demonstrate that NSG-MoE consistently surpasses strong baselines. Despite incorporating MoE -- which is typically computationally heavy -- our method achieves competitive training efficiency. Beyond empirical results, we provide a spectral analysis revealing that NSG performs adaptive filtering over modality-specific subspaces, thus explaining its disentangling behavior. Furthermore, an information-theoretic analysis shows that the architectural constraints imposed by NSG reduces mutual information between data and parameters and improving generalization capability. --- ### IntentCoding: Amplifying User Intent in Code Generation Authors: Zheng Fang, Yihong Dong, Lili Mou, Dongming Jin, Zhi Jin, Ge Li Published: 2026-01-20 Categories: cs.SE, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2602.00066v1 Large Language Models (LLMs) have shown strong capabilities in code generation, but their adherence to fine-grained user intent with multiple constraints remains a significant challenge. Our empirical analysis reveals two key observations: 1) Model performance deteriorates quickly as the number of constraints in the user intent increases, and 2) While user intent does influence the model's logits, such an influence may not be strong enough to effectively steer the decoding process. To this end, we propose Intent-Amplified Code Generation (IntentCoding), a novel decoding strategy that enhances an LLM's ability to follow user intent. IntentCoding captures the influence of user intent by masking out the intent, and applies a multi-strength ensemble mechanism to amplify the effect of user intent during generation. IntentCoding is model-agnostic, requires no additional training, and integrates seamlessly with existing decoding procedures. To enable systematic evaluation, we also construct CodeConstraints, a benchmark dataset specifically designed to test user intent compliance under varying numbers of constraints. Experiments on our constructed Constraints, as well as popular IFEvalCode, HumanEval and LiveCodeBench datasets, show that our IntentCoding model significantly improves both constraint satisfaction and functional correctness compared to standard decoding approaches. IntentCoding achieves up to 71.0% relative improvement on CodeConstraints, achieves up to 67.3% relative improvement on IFEvalCode and achieves up to 29.3% relative improvement in pass@1 on HumanEval and LiveCodeBench compared with greedy decoding. --- ### Responsible Evaluation of AI for Mental Health Authors: Hiba Arnaout, Anmol Goel, H. Andrew Schwartz, Steffen T. Eberhardt, Dana Atzil-Slonim, Gavin Doherty, Brian Schwartz, Wolfgang Lutz, Tim Althoff, Munmun De Choudhury, Hamidreza Jamalabadi, Raj Sanjay Shah et al. Published: 2026-01-20 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2602.00065v2 Although artificial intelligence (AI) shows growing promise for mental health care, current approaches to evaluating AI tools in this domain remain fragmented and poorly aligned with clinical practice, social context, and first-hand user experience. This paper argues for a rethinking of responsible evaluation -- what is measured, by whom, and for what purpose -- by introducing an interdisciplinary framework that integrates clinical soundness, social context, and equity, providing a structured basis for evaluation. Through an analysis of 135 recent *CL publications, we identify recurring limitations, including over-reliance on generic metrics that do not capture clinical validity, therapeutic appropriateness, or user experience, limited participation from mental health professionals, and insufficient attention to safety and equity. To address these gaps, we propose a taxonomy of AI mental health support types -- assessment-, intervention-, and information synthesis-oriented -- each with distinct risks and evaluative requirements, and illustrate its use through case studies. --- ### SPGCL: Simple yet Powerful Graph Contrastive Learning via SVD-Guided Structural Perturbation Authors: Hao Deng, Zhang Guo, Shuiping Gou, Bo Liu Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.00064v2 Graph Neural Networks (GNNs) are sensitive to structural noise from adversarial attacks or imperfections. Existing graph contrastive learning (GCL) methods typically rely on either random perturbations (e.g., edge dropping) for diversity or spectral augmentations (e.g., SVD) to preserve structural priors. However, random perturbations are structure-agnostic and may remove critical edges, while SVD-based views often lack sufficient diversity. Integrating these paradigms is challenging as they operate on discrete edge removal and continuous matrix factorization, respectively.We propose SPGCL, a framework for robust GCL via SVD-guided structural perturbation. Leveraging a recently developed SVD-based method that generalizes structural perturbation theory to arbitrary graphs, we design a two-stage strategy: (1) lightweight stochastic edge removal to inject diversity, and (2) truncated SVD to derive a structure-aware scoring matrix for sparse top-$P$ edge recovery. This integration offers three advantages: (1) Robustness to accidental deletion, as important edges can be recovered by SVD-guided scoring; (2) Enrichment with missing links, creating more informative contrastive views by introducing semantically meaningful edges; and (3) Controllable structural discrepancy, ensuring contrastive signals stem from semantic differences rather than edge-number gaps.Furthermore, we incorporate a contrastive fusion module with a global similarity constraint to align embeddings. Extensive experiments on ten benchmark datasets demonstrate that SPGCL consistently improves the robustness and accuracy of GNNs, outperforming state-of-the-art GCL and structure learning methods, validating its effectiveness in integrating previously disparate paradigms. --- ### The Impact of Machine Learning Uncertainty on the Robustness of Counterfactual Explanations Authors: Leonidas Christodoulou, Chang Sun Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.00063v1 Counterfactual explanations are widely used to interpret machine learning predictions by identifying minimal changes to input features that would alter a model's decision. However, most existing counterfactual methods have not been tested when model and data uncertainty change, resulting in explanations that may be unstable or invalid under real-world variability. In this work, we investigate the robustness of common combinations of machine learning models and counterfactual generation algorithms in the presence of both aleatoric and epistemic uncertainty. Through experiments on synthetic and real-world tabular datasets, we show that counterfactual explanations are highly sensitive to model uncertainty. In particular, we find that even small reductions in model accuracy - caused by increased noise or limited data - can lead to large variations in the generated counterfactuals on average and on individual instances. These findings underscore the need for uncertainty-aware explanation methods in domains such as finance and the social sciences. --- ### SCPL: Enhancing Neural Network Training Throughput with Decoupled Local Losses and Model Parallelism Authors: Ming-Yao Ho, Cheng-Kai Wang, You-Teng Lin, Hung-Hsuan Chen Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.00062v2 Adopting large-scale AI models in enterprise information systems is often hindered by high training costs and long development cycles, posing a significant managerial challenge. The standard end-to-end backpropagation (BP) algorithm is a primary driver of modern AI, but it is also the source of inefficiency in training deep networks. This paper introduces a new training methodology, Supervised Contrastive Parallel Learning (SCPL), that addresses this issue by decoupling BP and transforming a long gradient flow into multiple short ones. This design enables the simultaneous computation of parameter gradients in different layers, achieving superior model parallelism and enhancing training throughput. Detailed experiments are presented to demonstrate the efficiency and effectiveness of our model compared to BP, Early Exit, GPipe, and Associated Learning (AL), a state-of-the-art method for decoupling backpropagation. By mitigating a fundamental performance bottleneck, SCPL provides a practical pathway for organizations to develop and deploy advanced information systems more cost-effectively and with greater agility. The experimental code is released for reproducibility. https://github.com/minyaho/scpl/ --- ### Simple Role Assignment is Extraordinarily Effective for Safety Alignment Authors: Zhou Ziheng, Jiakun Ding, Zhaowei Zhang, Ruosen Gao, Yingnian Wu, Demetri Terzopoulos, Yipeng Kang, Fangwei Zhong, Junqi Wang Published: 2026-01-20 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2602.00061v1 Principle-based alignment often lacks context sensitivity and completeness. Grounded in Theory of Mind, we propose role conditioning as a compact alternative: social roles (e.g., mother, judge) implicitly encode both values and the cognitive schemas required to apply them. We introduce a training-free pipeline featuring a role-conditioned generator and iterative role-based critics for refinement. Across five model families, our approach consistently outperforms principle-based, Chain-of-Thought (CoT) and other baselines across benchmarks. Notably, it reduces unsafe outputs on the WildJailbreak benchmark from 81.4\% to 3.6\% with DeepSeek-V3. Not only for common safety benchmarks, it consistently applies for agentic safety tasks. These results establish role assignment as a powerful, interpretable paradigm for AI alignment and LLM-as-a-Judge construction. --- ### A longitudinal geospatial multimodal dataset of post-discharge frailty, physiology, mobility, and neighborhoods Authors: Ali Abedi, Charlene H. Chu, Shehroz S. Khan Published: 2026-01-20 Categories: cs.CY, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.00060v1 Frailty in older adults is associated with increased vulnerability to functional decline, reduced mobility, social isolation, and challenges during the transition from hospital to community living. These factors are associated with rehospitalization and may adversely influence recovery. Neighborhood environments can further shape recovery trajectories by affecting mobility opportunities, social engagement, and access to community resources. Multimodal sensing technologies combined with data-driven analytical approaches offer the potential to continuously monitor these multidimensional factors in real-world settings. This Data Descriptor presents GEOFRAIL, a longitudinal geospatial multimodal dataset collected from community-dwelling frail older adults following hospital discharge. The dataset is organized into interconnected tables capturing participant demographics, features derived from multimodal sensors, biweekly clinical assessments of frailty, physical function, and social isolation, and temporal location records linked to neighborhood amenities, crime rates, and census-based socioeconomic indicators. Data were collected over an eight-week post-discharge period using standardized pipelines with privacy-preserving spatial aggregation. Technical validation demonstrates internal consistency across geospatial, sensor-derived, and clinical measures and reports baseline performance of machine learning models for characterizing recovery trajectories. --- ### TextBFGS: A Case-Based Reasoning Approach to Code Optimization via Error-Operator Retrieval Authors: Zizheng Zhang, Yuyang Liao, Chen Chen, Jian He, Dun Wu, Qianjin Yu, Yanqin Gao, Jin Yang, Kailai Zhang, Eng Siong Chng, Xionghu Zhong Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2602.00059v2 Iterative code generation with Large Language Models (LLMs) can be viewed as an optimization process guided by textual feedback. However, existing LLM self-correction methods predominantly operate in a stateless, trial-and-error manner akin to first-order search, failing to leverage past problem-solving experiences. To bridge this gap, we introduce TextBFGS, a Case-Based Reasoning (CBR) framework inspired by the Quasi-Newton optimization method. Instead of retrieving raw, unstructured textual instances, TextBFGS maintains a dynamic Case Base of historical "Error-to-Operator" correction trajectories to approximate the semantic curvature (inverse Hessian matrix) of the task. Specifically, given a textual error feedback (the target problem), TextBFGS retrieves analogous historical correction patterns (Retrieve) and applies these abstract operators to refine the current code (Reuse/Revise). Furthermore, successful adaptations are continuously retained back into the Case Base (Retain), enabling a self-evolving system. Empirical evaluations on Python code optimization tasks (HumanEval, MBPP) demonstrate that TextBFGS significantly outperforms stateless baselines. It achieves superior pass rates with fewer model calls, establishing an efficient, experience-driven paradigm for LLM-based code optimization. --- ### Comparison of Multiple Classifiers for Android Malware Detection with Emphasis on Feature Insights Using CICMalDroid 2020 Dataset Authors: Md Min-Ha-Zul Abedin, Tazqia Mehrub Published: 2026-01-20 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2602.00058v1 Accurate Android malware detection was critical for protecting users at scale. Signature scanners lagged behind fast release cycles on public app stores. We aimed to build a trustworthy detector by pairing a comprehensive dataset with a rigorous, transparent evaluation, and to identify interpretable drivers of decisions. We used CICMalDroid2020, which contained 17,341 apps across Benign, Adware, Banking, SMS malware, and Riskware. We extracted 301 static and 263 dynamic features into a 564 dimensional hybrid vector, then evaluated seven classifiers under three schemes, original features, principal component analysis, PCA, and linear discriminant analysis, LDA, with a 70 percent training and 30 percent test split. Results showed that gradient boosting on the original features performed best. XGBoost achieved 0.9747 accuracy, 0.9703 precision, 0.9731 recall, and 0.9716 F1, and the confusion matrix indicated rare benign labels for malicious apps. HistGradientBoosting reached 0.9741 accuracy and 0.9708 F1, while CatBoost and Random Forest were slightly lower at 0.9678 and 0.9687 accuracy with 0.9636 and 0.9637 F1. KNN and SVM lagged. PCA reduced performance for all models, with XGBoost dropping to 0.9164 accuracy and 0.8988 F1. LDA maintained mid 90s accuracy and clarified separable clusters in projections. A depth two surrogate tree highlighted package name, main activity, and target SDK as key drivers. These findings established high fidelity supervised baselines for Android malware detection and indicated that rich hybrid features with gradient boosting offered a practical and interpretable foundation for deployment. --- ### Explore Brain-Inspired Machine Intelligence for Connecting Dots on Graphs Through Holographic Blueprint of Oscillatory Synchronization Authors: Tingting Dan, Jiaqi Ding, Guorong Wu Published: 2026-01-20 Categories: q-bio.NC, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.00057v1 Neural coupling in both neuroscience and artificial intelligence emerges as dynamic oscillatory patterns that encode abstract concepts. To this end, we hypothesize that a deeper understanding of the neural mechanisms governing brain rhythms can inspire next-generation design principles for machine learning algorithms, leading to improved efficiency and robustness. Building on this idea, we first model evolving brain rhythms through the interference of spontaneously synchronized neural oscillations, termed HoloBrain. The success of modeling brain rhythms using an artificial dynamical system of coupled oscillations motivates a "first principle" for brain-inspired machine intelligence based on a shared synchronization mechanism, termed HoloGraph. This principle enables graph neural networks to move beyond conventional heat diffusion paradigms toward modeling oscillatory synchronization. Our HoloGraph framework not only effectively mitigates the over-smoothing problem in graph neural networks but also demonstrates strong potential for reasoning and solving challenging problems on graphs. --- ### How Hyper-Datafication Impacts the Sustainability Costs in Frontier AI Authors: Sophia N. Wilson, Sebastian Mair, Mophat Okinyi, Erik B. Dam, Janin Koch, Raghavendra Selvan Published: 2026-01-20 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2602.00056v4 Large-scale data has fuelled the success of frontier artificial intelligence (AI) models over the past decade. This expansion has relied on sustained efforts by large technology corporations to aggregate and curate internet-scale datasets. In this work, we examine the environmental, social, and economic costs of large-scale data in AI through a sustainability lens. We argue that the field is shifting from building models from data to actively creating data for building models. We characterise this transition as hyper-datafication, which marks a critical juncture for the future of frontier AI and its societal impacts. To quantify and contextualise data-related costs, we analyse approximately 550,000 datasets from the Hugging Face Hub, focusing on dataset growth, storage-related energy consumption and carbon footprint, and societal representation using language data. We complement this analysis with qualitative responses from data workers in Kenya to examine the labour involved, including direct employment by big tech corporations and exposure to graphic content. We further draw on external data sources to substantiate our findings by illustrating the global disparity in data centre infrastructure. Our analyses reveal that hyper-datafication drives substantial and growing environmental costs while systematically redistributing labour risks and representational harms toward the Global South. Thus, we propose Data PROOFS recommendations spanning provenance, resource awareness, ownership, openness, frugality, and standards to mitigate these costs. Our work aims to make visible the often-overlooked costs of data that underpin frontier AI and to stimulate broader debate within the research community and beyond. --- ### BioNIC: Biologically Inspired Neural Network for Image Classification Using Connectomics Principles Authors: Diya Prasanth, Matthew Tivnan Published: 2026-01-20 Categories: cs.NE arXiv: https://arxiv.org/abs/2601.20876v1 We present BioNIC, a multi-layer feedforward neural network for emotion classification, inspired by detailed synaptic connectivity graphs from the MICrONs dataset. At a structural level, we incorporate architectural constraints derived from a single cortical column of the mouse Primary Visual Cortex(V1): connectivity imposed via adjacency masks, laminar organization, and graded inhibition representing inhibitory neurons. At the functional level, we implement biologically inspired learning: Hebbian synaptic plasticity with homeostatic regulation, Layer Normalization, data augmentation to model exposure to natural variability in sensory input, and synaptic noise to model neural stochasticity. We also include convolutional layers for spatial processing, mimicking retinotopic mapping. The model performance is evaluated on the Facial Emotion Recognition task FER-2013 and compared with a conventional baseline. Additionally, we investigate the impacts of each biological feature through a series of ablation experiments. While connectivity was limited to a single cortical column and biologically relevant connections, BioNIC achieved performance comparable to that of conventional models, with an accuracy of 59.77 $\pm$ 0.27% on FER-2013. Our findings demonstrate that integrating constraints derived from connectomics is a computationally plausible approach to developing biologically inspired artificial intelligence systems. This work also highlights the potential of new generation peta-scale connectomics data in advancing both neuroscience modeling and artificial intelligence. --- ### Variational Quantum Circuit-Based Reinforcement Learning for Dynamic Portfolio Optimization Authors: Vincent Gurgul, Ying Chen, Stefan Lessmann Published: 2026-01-20 Categories: cs.LG, q-fin.CP, q-fin.PM, quant-ph arXiv: https://arxiv.org/abs/2601.18811v2 This paper presents a Quantum Reinforcement Learning (QRL) solution to the dynamic portfolio optimization problem based on Variational Quantum Circuits. The implemented QRL approaches are quantum analogues of the classical neural-network-based Deep Deterministic Policy Gradient and Deep Q-Network algorithms. Through an empirical evaluation on real-world financial data, we show that our quantum agents achieve risk-adjusted performance comparable to, and in some cases exceeding, that of classical Deep RL models with several orders of magnitude more parameters. However, while quantum circuit execution is inherently fast at the hardware level, practical deployment on cloud-based quantum systems introduces substantial latency, making end-to-end runtime currently dominated by infrastructural overhead and limiting practical applicability. Taken together, our results suggest that QRL is theoretically competitive with state-of-the-art classical reinforcement learning and may become practically advantageous as deployment overheads diminish. This positions QRL as a promising paradigm for dynamic decision-making in complex, high-dimensional, and non-stationary environments such as financial markets. The complete codebase is released as open source at: https://github.com/VincentGurgul/qrl-dpo-public --- ### Atomic Depth Estimation From Noisy Electron Microscopy Data Via Deep Learning Authors: Matan Leibovich, Mai Tan, Ramon Manzorro, Adria Marcos-Morales, Sreyas Mohan, Peter A. Crozier, Carlos Fernandez-Granda Published: 2026-01-20 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.17046v3 We present a novel approach for extracting 3D atomic-level information from transmission electron microscopy (TEM) images affected by significant noise. The approach is based on formulating depth estimation as a semantic segmentation problem. We address the resulting segmentation problem by training a deep convolutional neural network to generate pixel-wise depth segmentation maps using simulated data corrupted by synthetic noise. The proposed method was applied to estimate the depth of atomic columns in CeO2 nanoparticles from simulated images and real-world TEM data. Our experiments show that the resulting depth estimates are accurate, calibrated and robust to noise. --- ### Interpretable and Sparse Linear Attention with Decoupled Membership-Subspace Modeling via MCR2 Objective Authors: Tianyuan Liu, Libin Hou, Linyuan Wang, Bin Yan Published: 2026-01-20 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.17042v1 Maximal Coding Rate Reduction (MCR2)-driven white-box transformer, grounded in structured representation learning, unifies interpretability and efficiency, providing a reliable white-box solution for visual modeling. However, in existing designs, tight coupling between "membership matrix" and "subspace matrix U" in MCR2 causes redundant coding under incorrect token projection. To this end, we decouple the functional relationship between the "membership matrix" and "subspaces U" in the MCR2 objective and derive an interpretable sparse linear attention operator from unrolled gradient descent of the optimized objective. Specifically, we propose to directly learn the membership matrix from inputs and subsequently derive sparse subspaces from the fullspace S. Consequently, gradient unrolling of the optimized MCR2 objective yields an interpretable sparse linear attention operator: Decoupled Membership-Subspace Attention (DMSA). Experimental results on visual tasks show that simply replacing the attention module in Token Statistics Transformer (ToST) with DMSA (we refer to as DMST) not only achieves a faster coding reduction rate but also outperforms ToST by 1.08%-1.45% in top-1 accuracy on the ImageNet-1K dataset. Compared with vanilla Transformer architectures, DMST exhibits significantly higher computational efficiency and interpretability. --- ### Arabic Sign Language Recognition using Multimodal Approach Authors: Ghadeer Alanazi, Abir Benabid Published: 2026-01-20 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.17041v1 Arabic Sign Language (ArSL) is an essential communication method for individuals in the Deaf and Hard-of-Hearing community. However, existing recognition systems face significant challenges due to their reliance on single sensor approaches like Leap Motion or RGB cameras. These systems struggle with limitations such as inadequate tracking of complex hand orientations and imprecise recognition of 3D hand movements. This research paper aims to investigate the potential of a multimodal approach that combines Leap Motion and RGB camera data to explore the feasibility of recognition of ArSL. The system architecture includes two parallel subnetworks: a custom dense neural network for Leap Motion data, incorporating dropout and L2 regularization, and an image subnetwork based on a fine-tuned VGG16 model enhanced with data augmentation techniques. Feature representations from both modalities are concatenated in a fusion model and passed through fully connected layers, with final classification performed via SoftMax activation to analyze spatial and temporal features of hand gestures. The system was evaluated on a custom dataset comprising 18 ArSL words, of which 13 were correctly recognized, yielding an overall accuracy of 78%. These results offer preliminary insights into the viability of multimodal fusion for sign language recognition and highlight areas for further optimization and dataset expansion. --- ### FP-THD: Full page transcription of historical documents Authors: H Neji, J Nogueras-Iso, J Lacasta, MÁ Latre, FJ García-Marco Published: 2026-01-20 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.17040v1 The transcription of historical documents written in Latin in XV and XVI centuries has special challenges as it must maintain the characters and special symbols that have distinct meanings to ensure that historical texts retain their original style and significance. This work proposes a pipeline for the transcription of historical documents preserving these special features. We propose to extend an existing text line recognition method with a layout analysis model. We analyze historical text images using a layout analysis model to extract text lines, which are then processed by an OCR model to generate a fully digitized page. We showed that our pipeline facilitates the processing of the page and produces an efficient result. We evaluated our approach on multiple datasets and demonstrate that the masked autoencoder effectively processes different types of text, including handwritten, printed and multi-language. --- ### MANGO: A Global Single-Date Paired Dataset for Mangrove Segmentation Authors: Junhyuk Heo, Beomkyu Choi, Hyunjin Shin, Darongsae Kwon Published: 2026-01-20 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.17039v2 Mangroves are critical for climate-change mitigation, requiring reliable monitoring for effective conservation. While deep learning has emerged as a powerful tool for mangrove detection, its progress is hindered by the limitations of existing datasets. In particular, many resources provide only annual map products without curated single-date image-mask pairs, limited to specific regions rather than global coverage, or remain inaccessible to the public. To address these challenges, we introduce MANGO, a large-scale global dataset comprising 42,703 labeled image-mask pairs across 124 countries. To construct this dataset, we retrieve all available Sentinel-2 imagery within the year 2020 for mangrove regions and select the best single-date observations that align with the mangrove annual mask. This selection is performed using a target detection-driven approach that leverages pixel-wise coordinate references to ensure adaptive and representative image-mask pairings. We also provide a benchmark across diverse semantic segmentation architectures under a country-disjoint split, establishing a foundation for scalable and reliable global mangrove monitoring. --- ### Hybrid Deep Feature Extraction and ML for Construction and Demolition Debris Classification Authors: Obai Alashram, Nejad Alagha, Mahmoud AlKakuri, Zeeshan Swaveel, Abigail Copiaco Published: 2026-01-20 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.17038v1 The construction industry produces significant volumes of debris, making effective sorting and classification critical for sustainable waste management and resource recovery. This study presents a hybrid vision-based pipeline that integrates deep feature extraction with classical machine learning (ML) classifiers for automated construction and demolition (C\&D) debris classification. A novel dataset comprising 1,800 balanced, high-quality images representing four material categories, Ceramic/Tile, Concrete, Trash/Waste, and Wood was collected from real construction sites in the UAE, capturing diverse real-world conditions. Deep features were extracted using a pre-trained Xception network, and multiple ML classifiers, including SVM, kNN, Bagged Trees, LDA, and Logistic Regression, were systematically evaluated. The results demonstrate that hybrid pipelines using Xception features with simple classifiers such as Linear SVM, kNN, and Bagged Trees achieve state-of-the-art performance, with up to 99.5\% accuracy and macro-F1 scores, surpassing more complex or end-to-end deep learning approaches. The analysis highlights the operational benefits of this approach for robust, field-deployable debris identification and provides pathways for future integration with robotics and onsite automation systems. --- ### AMVICC: A Novel Benchmark for Cross-Modal Failure Mode Profiling for VLMs and IGMs Authors: Aahana Basappa, Pranay Goel, Anusri Karra, Anish Karra, Asa Gilmore, Kevin Zhu Published: 2026-01-20 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.17037v2 We investigate visual reasoning limitations of both multimodal large language models (MLLMs) and image generation models (IGMs) by creating a novel benchmark to systematically compare failure modes across image-to-text and text-to-image tasks, enabling cross-modal evaluation of visual understanding. Despite rapid growth in machine learning, vision language models (VLMs) still fail to understand basic visual concepts such as object orientation, quantity, and spatial relationships, which highlights gaps in elementary visual reasoning. By adapting MMVP benchmark questions into explicit and implicit prompts, we create \textit{AMVICC}, a novel benchmark for profiling failure modes across various modalities. After testing 11 MLLMs and 3 IGMs in 9 categories of visual reasoning, our results show that failure modes are often shared between models and modalities. However, certain failures are model-specific and modality-specific, and this can potentially be attributed to various factors. IGMs consistently struggle to manipulate specific visual components in response to prompts, especially in explicit prompts, suggesting poor control over fine-grained visual attributes. Our findings apply most directly to the evaluation of existing state-of-the-art models on structured visual reasoning tasks. This work lays the foundation for future cross-modal alignment studies, offering a framework to probe whether image generation and visual interpretation failures stem from shared limitations. These insights can guide future improvements in unified vision-language modeling. --- ### Policy-Embedded Graph Expansion: Networked HIV Testing with Diffusion-Driven Network Samples Authors: Akseli Kangaslahti, Davin Choo, Lingkai Kong, Milind Tambe, Alastair van Heerden, Cheryl Johnson Published: 2026-01-20 Categories: cs.SI, cs.AI arXiv: https://arxiv.org/abs/2601.16233v2 HIV is a retrovirus that attacks the human immune system and can lead to death without proper treatment. In collaboration with the WHO and the University of Witwatersrand, we study how to improve the efficiency of HIV testing with the goal of eventual deployment, directly supporting progress toward UN Sustainable Development Goal 3.3. While prior work has demonstrated the promise of intelligent algorithms for sequential, network-based HIV testing, existing approaches rely on assumptions that are impractical in our real-world implementations. Here, we study sequential testing on incrementally revealed disease networks and introduce Policy-Embedded Graph Expansion (PEGE), a novel framework that directly embeds a generative distribution over graph expansions into the decision-making policy rather than attempting explicit topological reconstruction. We further propose Dynamics-Driven Branching (DDB), a diffusion-based graph expansion model that supports decision making in PEGE and is designed for data-limited settings where forest structures arise naturally, as in our real-world referral process. Experiments on real HIV transmission networks show that the combined approach (PEGE + DDB) consistently outperforms baselines (e.g., 17.3% improvement in discounted reward and 15.4% more HIV detections with 25% of the population tested) and explore key tradeoffs that drive solution quality. --- ### SoundBreak: A Systematic Study of Audio-Only Adversarial Attacks on Trimodal Models Authors: Aafiya Hussain, Gaurav Srivastava, Alvi Ishmam, Zaber Hakim, Chris Thomas Published: 2026-01-20 Categories: cs.SD, cs.AI, cs.CL, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2601.16231v1 Multimodal foundation models that integrate audio, vision, and language achieve strong performance on reasoning and generation tasks, yet their robustness to adversarial manipulation remains poorly understood. We study a realistic and underexplored threat model: untargeted, audio-only adversarial attacks on trimodal audio-video-language models. We analyze six complementary attack objectives that target different stages of multimodal processing, including audio encoder representations, cross-modal attention, hidden states, and output likelihoods. Across three state-of-the-art models and multiple benchmarks, we show that audio-only perturbations can induce severe multimodal failures, achieving up to 96% attack success rate. We further show that attacks can be successful at low perceptual distortions (LPIPS <= 0.08, SI-SNR >= 0) and benefit more from extended optimization than increased data scale. Transferability across models and encoders remains limited, while speech recognition systems such as Whisper primarily respond to perturbation magnitude, achieving >97% attack success under severe distortion. These results expose a previously overlooked single-modality attack surface in multimodal systems and motivate defenses that enforce cross-modal consistency. --- ### Zero-Shot Speech LLMs for Multi-Aspect Evaluation of L2 Speech: Challenges and Opportunities Authors: Aditya Kamlesh Parikh, Cristian Tejedor-Garcia, Catia Cucchiarini, Helmer Strik Published: 2026-01-20 Categories: eess.AS, cs.AI, cs.CL, cs.SD arXiv: https://arxiv.org/abs/2601.16230v1 An accurate assessment of L2 English pronunciation is crucial for language learning, as it provides personalized feedback and ensures a fair evaluation of individual progress. However, automated scoring remains challenging due to the complexity of sentence-level fluency, prosody, and completeness. This paper evaluates the zero-shot performance of Qwen2-Audio-7B-Instruct, an instruction-tuned speech-LLM, on 5,000 Speechocean762 utterances. The model generates rubric-aligned scores for accuracy, fluency, prosody, and completeness, showing strong agreement with human ratings within +-2 tolerance, especially for high-quality speech. However, it tends to overpredict low-quality speech scores and lacks precision in error detection. These findings demonstrate the strong potential of speech LLMs in scalable pronunciation assessment and suggest future improvements through enhanced prompting, calibration, and phonetic integration to advance Computer-Assisted Pronunciation Training. --- ### Latent Causal Diffusions for Single-Cell Perturbation Modeling Authors: Lars Lorch, Jiaqi Zhang, Charlotte Bunne, Andreas Krause, Bernhard Schölkopf, Caroline Uhler Published: 2026-01-20 Categories: q-bio.MN, cs.LG arXiv: https://arxiv.org/abs/2601.15341v1 Perturbation screens hold the potential to systematically map regulatory processes at single-cell resolution, yet modeling and predicting transcriptome-wide responses to perturbations remains a major computational challenge. Existing methods often underperform simple baselines, fail to disentangle measurement noise from biological signal, and provide limited insight into the causal structure governing cellular responses. Here, we present the latent causal diffusion (LCD), a generative model that frames single-cell gene expression as a stationary diffusion process observed under measurement noise. LCD outperforms established approaches in predicting the distributional shifts of unseen perturbation combinations in single-cell RNA-sequencing screens while simultaneously learning a mechanistic dynamical system of gene regulation. To interpret these learned dynamics, we develop an approach we call causal linearization via perturbation responses (CLIPR), which yields an approximation of the direct causal effects between all genes modeled by the diffusion. CLIPR provably identifies causal effects under a linear drift assumption and recovers causal structure in both simulated systems and a genome-wide perturbation screen, where it clusters genes into coherent functional modules and resolves causal relationships that standard differential expression analysis cannot. The LCD-CLIPR framework bridges generative modeling with causal inference to predict unseen perturbation effects and map the underlying regulatory mechanisms of the transcriptome. --- ### Learning Nonlinear Heterogeneity in Physical Kolmogorov-Arnold Networks Authors: Fabiana Taglietti, Andrea Pulici, Maxwell Roxburgh, Gabriele Seguini, Ian Vidamour, Stephan Menzel, Edoardo Franco, Michele Laus, Eleni Vasilaki, Michele Perego, Thomas J. Hayward, Marco Fanciulli et al. Published: 2026-01-20 Categories: cond-mat.dis-nn, cond-mat.mes-hall, cs.LG, nlin.AO, physics.app-ph arXiv: https://arxiv.org/abs/2601.15340v3 Physical neural networks typically train linear synaptic weights while treating device nonlinearities as fixed. We show the opposite - by training the synaptic nonlinearity itself, as in Kolmogorov-Arnold Network (KAN) architectures, we yield markedly higher task performance per physical resource and improved performance-parameter scaling than conventional linear weight-based networks, demonstrating ability of KAN topologies to exploit reconfigurable nonlinear physical dynamics. We experimentally realise physical KANs in silicon-on-insulator devices we term 'Synaptic Nonlinear Elements' (SYNEs), operating at room temperature, microampere currents, 2 MHz speeds and ~750 fJ per nonlinear operation, with no observed degradation over 10^13 measurements and months-long timescales. We demonstrate nonlinear function regression, classification, and prediction of Li-Ion battery dynamics from noisy real-world multi-sensor data. Physical KANs outperform equivalently-parameterised software multilayer perceptron networks across all tasks, with up to two orders of magnitude fewer parameters, and two orders of magnitude fewer devices than linear weight based physical networks. These results establish learned physical nonlinearity as a hardware-native computational primitive for compact and efficient learning systems, and SYNE devices as effective substrates for heterogenous nonlinear computing. --- ### Lost in Transcription: How Speech-to-Text Errors Derail Code Understanding Authors: Jayant Havare, Ashish Mittal, Srikanth Tamilselvam, Ganesh Ramakrishnan Published: 2026-01-20 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.15339v1 Code understanding is a foundational capability in software engineering tools and developer workflows. However, most existing systems are designed for English-speaking users interacting via keyboards, which limits accessibility in multilingual and voice-first settings, particularly in regions like India. Voice-based interfaces offer a more inclusive modality, but spoken queries involving code present unique challenges due to the presence of non-standard English usage, domain-specific vocabulary, and custom identifiers such as variable and function names, often combined with code-mixed expressions. In this work, we develop a multilingual speech-driven framework for code understanding that accepts spoken queries in a user native language, transcribes them using Automatic Speech Recognition (ASR), applies code-aware ASR output refinement using Large Language Models (LLMs), and interfaces with code models to perform tasks such as code question answering and code retrieval through benchmarks such as CodeSearchNet, CoRNStack, and CodeQA. Focusing on four widely spoken Indic languages and English, we systematically characterize how transcription errors impact downstream task performance. We also identified key failure modes in ASR for code and demonstrated that LLM-guided refinement significantly improves performance across both transcription and code understanding stages. Our findings underscore the need for code-sensitive adaptations in speech interfaces and offer a practical solution for building robust, multilingual voice-driven programming tools. --- ### From Quotes to Concepts: Axial Coding of Political Debates with Ensemble LMs Authors: Angelina Parfenova, David Graus, Juergen Pfeffer Published: 2026-01-20 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.15338v1 Axial coding is a commonly used qualitative analysis method that enhances document understanding by organizing sentence-level open codes into broader categories. In this paper, we operationalize axial coding with large language models (LLMs). Extending an ensemble-based open coding approach with an LLM moderator, we add an axial coding step that groups open codes into higher-order categories, transforming raw debate transcripts into concise, hierarchical representations. We compare two strategies: (i) clustering embeddings of code-utterance pairs using density-based and partitioning algorithms followed by LLM labeling, and (ii) direct LLM-based grouping of codes and utterances into categories. We apply our method to Dutch parliamentary debates, converting lengthy transcripts into compact, hierarchically structured codes and categories. We evaluate our method using extrinsic metrics aligned with human-assigned topic labels (ROUGE-L, cosine, BERTScore), and intrinsic metrics describing code groups (coverage, brevity, coherence, novelty, JSD divergence). Our results reveal a trade-off: density-based clustering achieves high coverage and strong cluster alignment, while direct LLM grouping results in higher fine-grained alignment, but lower coverage 20%. Overall, clustering maximizes coverage and structural separation, whereas LLM grouping produces more concise, interpretable, and semantically aligned categories. To support future research, we publicly release the full dataset of utterances and codes, enabling reproducibility and comparative studies. --- ### Language Models Entangle Language and Culture Authors: Shourya Jain, Paras Chopra Published: 2026-01-20 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2601.15337v1 Users should not be systemically disadvantaged by the language they use for interacting with LLMs; i.e. users across languages should get responses of similar quality irrespective of language used. In this work, we create a set of real-world open-ended questions based on our analysis of the WildChat dataset and use it to evaluate whether responses vary by language, specifically, whether answer quality depends on the language used to query the model. We also investigate how language and culture are entangled in LLMs such that choice of language changes the cultural information and context used in the response by using LLM-as-a-Judge to identify the cultural context present in responses. To further investigate this, we evaluate LLMs on a translated subset of the CulturalBench benchmark across multiple languages. Our evaluations reveal that LLMs consistently provide lower quality answers to open-ended questions in low resource languages. We find that language significantly impacts the cultural context used by the model. This difference in context impacts the quality of the downstream answer. --- ### Learning Discrete Successor Transitions in Continuous Attractor Networks: Emergence, Limits, and Topological Constraints Authors: Daniel Brownell Published: 2026-01-20 Categories: q-bio.NC, cs.AI arXiv: https://arxiv.org/abs/2601.15336v1 Continuous attractor networks (CANs) are a well-established class of models for representing low-dimensional continuous variables such as head direction, spatial position, and phase. In canonical spatial domains, transitions along the attractor manifold are driven by continuous displacement signals, such as angular velocity-provided by sensorimotor systems external to the CAN itself. When such signals are not explicitly provided as dedicated displacement inputs, it remains unclear whether attractor-based circuits can reliably acquire recurrent dynamics that support stable state transitions, or whether alternative predictive strategies dominate. In this work, we present an experimental framework for training CANs to perform successor-like transitions between stable attractor states in the absence of externally provided displacement signals. We compare two recurrent topologies, a circular ring and a folded snake manifold, and systematically vary the temporal regime under which stability is evaluated. We find that, under short evaluation windows, networks consistently converge to impulse-driven associative solutions that achieve high apparent accuracy yet lack persistent attractor dynamics. Only when stability is explicitly enforced over extended free-run periods do genuine attractor-based transition dynamics emerge. This suggests that shortcut solutions are the default outcome of local learning in recurrent networks, while attractor dynamics represent a constrained regime rather than a generic result. Furthermore, we demonstrate that topology strictly limits the capacity for learned transitions. While the continuous ring topology achieves perfect stability over long horizons, the folded snake topology hits a geometric limit characterized by failure at manifold discontinuities, which neither curriculum learning nor basal ganglia-inspired gating can fully overcome. --- ### ToolCaching: Towards Efficient Caching for LLM Tool-calling Authors: Yi Zhai, Dian Shen, Junzhou Luo, Bin Yang Published: 2026-01-20 Categories: cs.SE, cs.AI, cs.PL arXiv: https://arxiv.org/abs/2601.15335v1 Recent advances in Large Language Models (LLMs) have revolutionized web applications, enabling intelligent search, recommendation, and assistant services with natural language interfaces. Tool-calling extends LLMs with the ability to interact with external APIs, greatly enhancing their practical utility. While prior research has improved tool-calling performance by adopting traditional computer systems techniques, such as parallel and asynchronous execution, the challenge of redundant or repeated tool-calling requests remains largely unaddressed. Caching is a classic solution to this problem, but applying it to LLM tool-calling introduces new difficulties due to heterogeneous request semantics, dynamic workloads, and varying freshness requirements, which render conventional cache policies ineffective. To address these issues, we propose ToolCaching, an efficient feature-driven and adaptive caching framework for LLM tool-calling systems. ToolCaching systematically integrates semantic and system-level features to evaluate request cacheability and estimate caching value. At its core, the VAAC algorithm integrates bandit-based admission with value-driven, multi-factor eviction, jointly accounting for request frequency, recency, and caching value. Extensive experiments on synthetic and public tool-calling workloads demonstrate that ToolCaching with VAAC achieves up to 11% higher cache hit ratios and 34% lower latency compared to standard policies, effectively accelerating LLM tool-calling in practical applications. --- ### No Reliable Evidence of Self-Reported Sentience in Small Large Language Models Authors: Caspar Kaiser, Sean Enderby Published: 2026-01-20 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.15334v2 Whether language models possess sentience has no empirical answer. But whether they believe themselves to be sentient can, in principle, be tested. We do so by querying several open-weights models about their own consciousness, and then verifying their responses using classifiers trained on internal activations. We draw upon three model families (Qwen, Llama, GPT-OSS) ranging from 0.6 billion to 70 billion parameters, approximately 50 questions about consciousness and subjective experience, and three classification methods from the interpretability literature. First, we find that models consistently deny being sentient: they attribute consciousness to humans but not to themselves. Second, classifiers trained to detect underlying beliefs - rather than mere outputs - provide no clear evidence that these denials are untruthful. Third, within the Qwen family, larger models deny sentience more confidently than smaller ones. These findings contrast with recent work suggesting that models harbour latent beliefs in their own consciousness. --- ### Empowering LLMs for Structure-Based Drug Design via Exploration-Augmented Latent Inference Authors: Xuanning Hu, Anchen Li, Qianli Xing, Jinglong Ji, Hao Tuo, Bo Yang Published: 2026-01-20 Categories: cs.LG, cs.AI, q-bio.QM arXiv: https://arxiv.org/abs/2601.15333v2 Large Language Models (LLMs) possess strong representation and reasoning capabilities, but their application to structure-based drug design (SBDD) is limited by insufficient understanding of protein structures and unpredictable molecular generation. To address these challenges, we propose Exploration-Augmented Latent Inference for LLMs (ELILLM), a framework that reinterprets the LLM generation process as an encoding, latent space exploration, and decoding workflow. ELILLM explicitly explores portions of the design problem beyond the model's current knowledge while using a decoding module to handle familiar regions, generating chemically valid and synthetically reasonable molecules. In our implementation, Bayesian optimization guides the systematic exploration of latent embeddings, and a position-aware surrogate model efficiently predicts binding affinity distributions to inform the search. Knowledge-guided decoding further reduces randomness and effectively imposes chemical validity constraints. We demonstrate ELILLM on the CrossDocked2020 benchmark, showing strong controlled exploration and high binding affinity scores compared with seven baseline methods. These results demonstrate that ELILLM can effectively enhance LLMs capabilities for SBDD. --- ### RECAP: A Resource-Efficient Method for Adversarial Prompting in Large Language Models Authors: Rishit Chugh Published: 2026-01-20 Categories: cs.CL, cs.AI, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2601.15331v1 The deployment of large language models (LLMs) has raised security concerns due to their susceptibility to producing harmful or policy-violating outputs when exposed to adversarial prompts. While alignment and guardrails mitigate common misuse, they remain vulnerable to automated jailbreaking methods such as GCG, PEZ, and GBDA, which generate adversarial suffixes via training and gradient-based search. Although effective, these methods particularly GCG are computationally expensive, limiting their practicality for organisations with constrained resources. This paper introduces a resource-efficient adversarial prompting approach that eliminates the need for retraining by matching new prompts to a database of pre-trained adversarial prompts. A dataset of 1,000 prompts was classified into seven harm-related categories, and GCG, PEZ, and GBDA were evaluated on a Llama 3 8B model to identify the most effective attack method per category. Results reveal a correlation between prompt type and algorithm effectiveness. By retrieving semantically similar successful adversarial prompts, the proposed method achieves competitive attack success rates with significantly reduced computational cost. This work provides a practical framework for scalable red-teaming and security evaluation of aligned LLMs, including in settings where model internals are inaccessible. --- ### ICPO: Illocution-Calibrated Policy Optimization for Multi-Turn Conversation Authors: Zhebo Wang, Xiaohu Mu, Zijie Zhou, Mohan Li, Wenpeng Xing, Dezhang Kong, Meng Han Published: 2026-01-20 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.15330v1 Large Language Models (LLMs) in multi-turn conversations often suffer from a ``lost-in-conversation'' phenomenon, where they struggle to recover from early incorrect assumptions, particularly when users provide ambiguous initial instructions. We find that standard post-training techniques like Reinforcement Learning with Verifiable Rewards (RLVR) exacerbate this issue by rewarding confident, direct answers, thereby inducing overconfidence and discouraging the model from seeking clarification. To address this, we propose Illocution-Calibrated Policy Optimization (ICPO), a novel training framework that sensitizes the model to instruction ambiguity. ICPO augments the training corpus with underspecified prompts and conditions the reward signal on the user's illocutionary intent, rewarding the model for expressing uncertainty or asking for clarification when faced with ambiguity. Experiments demonstrate that ICPO fosters appropriate humility, yielding a substantial average improvement of 75\% in multi-turn conversation, while preserving robust performance on single-turn benchmarks. Our work presents a practical path toward more robust and collaborative conversational AI that can better navigate the nuances of human interaction. --- ### Report for NSF Workshop on AI for Electronic Design Automation Authors: Deming Chen, Vijay Ganesh, Weikai Li, Yingyan Celine Lin, Yong Liu, Subhasish Mitra, David Z. Pan, Ruchir Puri, Jason Cong, Yizhou Sun Published: 2026-01-20 Categories: cs.LG, cs.AI, cs.AR arXiv: https://arxiv.org/abs/2601.14541v4 This report distills the discussions and recommendations from the NSF Workshop on AI for Electronic Design Automation (EDA), held on December 10, 2024 in Vancouver alongside NeurIPS 2024. Bringing together experts across machine learning and EDA, the workshop examined how AI-spanning large language models (LLMs), graph neural networks (GNNs), reinforcement learning (RL), neurosymbolic methods, etc.-can facilitate EDA and shorten design turnaround. The workshop includes four themes: (1) AI for physical synthesis and design for manufacturing (DFM), discussing challenges in physical manufacturing process and potential AI applications; (2) AI for high-level and logic-level synthesis (HLS/LLS), covering pragma insertion, program transformation, RTL code generation, etc.; (3) AI toolbox for optimization and design, discussing frontier AI developments that could potentially be applied to EDA tasks; and (4) AI for test and verification, including LLM-assisted verification tools, ML-augmented SAT solving, security/reliability challenges, etc. The report recommends NSF to foster AI/EDA collaboration, invest in foundational AI for EDA, develop robust data infrastructures, promote scalable compute infrastructure, and invest in workforce development to democratize hardware design and enable next-generation hardware systems. The workshop information can be found on the website https://ai4eda-workshop.github.io/. --- ### engGNN: A Dual-Graph Neural Network for Omics-Based Disease Classification and Feature Selection Authors: Tiantian Yang, Yuxuan Wang, Zhenwei Zhou, Ching-Ti Liu Published: 2026-01-20 Categories: cs.LG, q-bio.GN, stat.ML arXiv: https://arxiv.org/abs/2601.14536v1 Omics data, such as transcriptomics, proteomics, and metabolomics, provide critical insights into disease mechanisms and clinical outcomes. However, their high dimensionality, small sample sizes, and intricate biological networks pose major challenges for reliable prediction and meaningful interpretation. Graph Neural Networks (GNNs) offer a promising way to integrate prior knowledge by encoding feature relationships as graphs. Yet, existing methods typically rely solely on either an externally curated feature graph or a data-driven generated one, which limits their ability to capture complementary information. To address this, we propose the external and generated Graph Neural Network (engGNN), a dual-graph framework that jointly leverages both external known biological networks and data-driven generated graphs. Specifically, engGNN constructs a biologically informed undirected feature graph from established network databases and complements it with a directed feature graph derived from tree-ensemble models. This dual-graph design produces more comprehensive embeddings, thereby improving predictive performance and interpretability. Through extensive simulations and real-world applications to gene expression data, engGNN consistently outperforms state-of-the-art baselines. Beyond classification, engGNN provides interpretable feature importance scores that facilitate biologically meaningful discoveries, such as pathway enrichment analysis. Taken together, these results highlight engGNN as a robust, flexible, and interpretable framework for disease classification and biomarker discovery in high-dimensional omics contexts. --- ### Search over Self-Edit Strategies for LLM Adaptation Authors: Alistair Cheong, Haolin Cong, Tyler Yang, Dustin Miao Published: 2026-01-20 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.14532v1 Many LLM-based open-ended search systems freeze the foundation model that proposes improvements to existing solutions, which may bottleneck long-run progress. Recent work has explored updating the proposal model at test time (arXiv:2511.23473), but the update strategy is still typically hand-specified. Therefore, this study investigated whether an LLM can use task feedback to decide how it should update its weights. For tractability, we focused on the simpler case where there is only one round of self-improvement, and restricted the update operator to self-supervised next token prediction (NTP), leaving the model freedom in choosing its training data and key NTP hyperparameters. Using the Self-Adapting Language Models (SEAL) (arXiv:2506.10943) framework as a testbed, we relaxed its fixed human template constraint and allowed the model to generate its own self-edit templates, thereby giving it more control over its training data and hyperparameters. Two variants were studied, differing in whether template generation was conditioned on a lightweight archive of past templates. In SEAL's Single-Passage Knowledge Incorporation setting with Qwen3-8B on SQuAD (arXiv:1606.05250), the no-archive variant performed comparably to the weaker "Implications" baseline, while the archive variant outperformed "Implications" and approached the strongest human-designed "Rewrite" baseline without surpassing it. Further analysis of collapse in the model's exploration revealed that a naive archive can confer some short-term robustness but can also accelerate homogenization, suggesting that explicit novelty pressure may be required to consistently advance beyond carefully optimized human strategies. Our code is available at https://github.com/cheongalc/search-self-edit-strategies . --- ### PAS-Mamba: Phase-Amplitude-Spatial State Space Model for MRI Reconstruction Authors: Xiaoyan Kui, Zijie Fan, Zexin Ji, Qinsong Li, Hao Xu, Weixin Si, Haodong Xu, Beiji Zou Published: 2026-01-20 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.14530v1 Joint feature modeling in both the spatial and frequency domains has become a mainstream approach in MRI reconstruction. However, existing methods generally treat the frequency domain as a whole, neglecting the differences in the information carried by its internal components. According to Fourier transform theory, phase and amplitude represent different types of information in the image. Our spectrum swapping experiments show that magnitude mainly reflects pixel-level intensity, while phase predominantly governs image structure. To prevent interference between phase and magnitude feature learning caused by unified frequency-domain modeling, we propose the Phase-Amplitude-Spatial State Space Model (PAS-Mamba) for MRI Reconstruction, a framework that decouples phase and magnitude modeling in the frequency domain and combines it with image-domain features for better reconstruction. In the image domain, LocalMamba preserves spatial locality to sharpen fine anatomical details. In frequency domain, we disentangle amplitude and phase into two specialized branches to avoid representational coupling. To respect the concentric geometry of frequency information, we propose Circular Frequency Domain Scanning (CFDS) to serialize features from low to high frequencies. Finally, a Dual-Domain Complementary Fusion Module (DDCFM) adaptively fuses amplitude phase representations and enables bidirectional exchange between frequency and image domains, delivering superior reconstruction. Extensive experiments on the IXI and fastMRI knee datasets show that PAS-Mamba consistently outperforms state of the art reconstruction methods. --- ### LLM Security and Safety: Insights from Homotopy-Inspired Prompt Obfuscation Authors: Luis Lazo, Hamed Jelodar, Roozbeh Razavi-Far Published: 2026-01-20 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2601.14528v1 In this study, we propose a homotopy-inspired prompt obfuscation framework to enhance understanding of security and safety vulnerabilities in Large Language Models (LLMs). By systematically applying carefully engineered prompts, we demonstrate how latent model behaviors can be influenced in unexpected ways. Our experiments encompassed 15,732 prompts, including 10,000 high-priority cases, across LLama, Deepseek, KIMI for code generation, and Claude to verify. The results reveal critical insights into current LLM safeguards, highlighting the need for more robust defense mechanisms, reliable detection strategies, and improved resilience. Importantly, this work provides a principled framework for analyzing and mitigating potential weaknesses, with the goal of advancing safe, responsible, and trustworthy AI technologies. --- ### Towards Execution-Grounded Automated AI Research Authors: Chenglei Si, Zitong Yang, Yejin Choi, Emmanuel Candès, Diyi Yang, Tatsunori Hashimoto Published: 2026-01-20 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.14525v1 Automated AI research holds great potential to accelerate scientific discovery. However, current LLMs often generate plausible-looking but ineffective ideas. Execution grounding may help, but it is unclear whether automated execution is feasible and whether LLMs can learn from the execution feedback. To investigate these, we first build an automated executor to implement ideas and launch large-scale parallel GPU experiments to verify their effectiveness. We then convert two realistic research problems - LLM pre-training and post-training - into execution environments and demonstrate that our automated executor can implement a large fraction of the ideas sampled from frontier LLMs. We analyze two methods to learn from the execution feedback: evolutionary search and reinforcement learning. Execution-guided evolutionary search is sample-efficient: it finds a method that significantly outperforms the GRPO baseline (69.4% vs 48.0%) on post-training, and finds a pre-training recipe that outperforms the nanoGPT baseline (19.7 minutes vs 35.9 minutes) on pre-training, all within just ten search epochs. Frontier LLMs often generate meaningful algorithmic ideas during search, but they tend to saturate early and only occasionally exhibit scaling trends. Reinforcement learning from execution reward, on the other hand, suffers from mode collapse. It successfully improves the average reward of the ideator model but not the upper-bound, due to models converging on simple ideas. We thoroughly analyze the executed ideas and training dynamics to facilitate future efforts towards execution-grounded automated AI research. --- ### Large Language Model-Powered Evolutionary Code Optimization on a Phylogenetic Tree Authors: Leyi Zhao, Weijie Huang, Yitong Guo, Jiang Bian, Chenghong Wang, Xuhong Zhang Published: 2026-01-20 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.14523v1 Optimizing scientific computing algorithms for modern GPUs is a labor-intensive and iterative process involving repeated code modification, benchmarking, and tuning across complex hardware and software stacks. Recent work has explored large language model (LLM)-assisted evolutionary methods for automated code optimization, but these approaches primarily rely on outcome-based selection and random mutation, underutilizing the rich trajectory information generated during iterative optimization. We propose PhyloEvolve, an LLM-agent system that reframes GPU-oriented algorithm optimization as an In-Context Reinforcement Learning (ICRL) problem. This formulation enables trajectory-conditioned reuse of optimization experience without model retraining. PhyloEvolve integrates Algorithm Distillation and prompt-based Decision Transformers into an iterative workflow, treating sequences of algorithm modifications and performance feedback as first-class learning signals. To organize optimization history, we introduce a phylogenetic tree representation that captures inheritance, divergence, and recombination among algorithm variants, enabling backtracking, cross-lineage transfer, and reproducibility. The system combines elite trajectory pooling, multi-island parallel exploration, and containerized execution to balance exploration and exploitation across heterogeneous hardware. We evaluate PhyloEvolve on scientific computing workloads including PDE solvers, manifold learning, and spectral graph algorithms, demonstrating consistent improvements in runtime, memory efficiency, and correctness over baseline and evolutionary methods. Code is published at: https://github.com/annihi1ation/phylo_evolve --- ### On the Runway Cascade of Transformers for Language Modeling Authors: Hunjae Lee, Corey Clark Published: 2026-01-20 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.14522v1 In decoder-only (causal) transformers, the computation graph created by causal masking routes information through both direct-path attention and indirect paths formed by intermediate tokens. We denote these indirect paths between token pairs as their runways. We argue that certain failure modes of causal transformers as observed by a growing body of recent works are likely exacerbated by a misalignment between these two information propagation modes. We formalize runway cascade as a phenomenon whereby this misalignment results in redundancies and irrelevant information cascading to token representations despite adequately learned attention patterns. As a solution, we propose runway-aware rewiring as a more explicit way of incorporating runway context directly into each token's direct-path attention. This mechanism re-wires the attention pattern for each token based on a summary of its runway landscape, enabling awareness of accumulating representational influences and allowing for more balanced information propagation. Our proposed methodology introduces no additional parameters and can seamlessly be integrated into standard attention mechanism. Empirically, our rewired transformer results in steady improvements in general language modeling as well as noticeably stronger information retrieval and extrapolation abilities compared to standard transformers. --- ### How Worst-Case Are Adversarial Attacks? Linking Adversarial and Perturbation Robustness Authors: Giulio Rossolini Published: 2026-01-20 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.14519v2 Adversarial attacks are widely used to identify model vulnerabilities; however, their validity as proxies for robustness to random perturbations remains debated. We ask whether an adversarial example provides a representative estimate of misprediction risk under stochastic perturbations of the same magnitude, or instead reflects an atypical worst-case event. To address this question, we introduce a probabilistic analysis that quantifies this risk with respect to directionally biased perturbation distributions, parameterized by a concentration factor $κ$ that interpolates between isotropic noise and adversarial directions. Building on this, we study the limits of this connection by proposing an attack strategy designed to probe vulnerabilities in regimes that are statistically closer to uniform noise. Experiments on ImageNet and CIFAR-10 systematically benchmark multiple attacks, revealing when adversarial success meaningfully reflects robustness to perturbations and when it does not, thereby informing their use in safety-oriented robustness evaluation. --- ### Business Logic-Driven Text-to-SQL Data Synthesis for Business Intelligence Authors: Jinhui Liu, Ximeng Zhang, Yanbo Ai, Zhou Yu Published: 2026-01-20 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.14518v1 Evaluating Text-to-SQL agents in private business intelligence (BI) settings is challenging due to the scarcity of realistic, domain-specific data. While synthetic evaluation data offers a scalable solution, existing generation methods fail to capture business realism--whether questions reflect realistic business logic and workflows. We propose a Business Logic-Driven Data Synthesis framework that generates data grounded in business personas, work scenarios, and workflows. In addition, we improve the data quality by imposing a business reasoning complexity control strategy that diversifies the analytical reasoning steps required to answer the questions. Experiments on a production-scale Salesforce database show that our synthesized data achieves high business realism (98.44%), substantially outperforming OmniSQL (+19.5%) and SQL-Factory (+54.7%), while maintaining strong question-SQL alignment (98.59%). Our synthetic data also reveals that state-of-the-art Text-to-SQL models still have significant performance gaps, achieving only 42.86% execution accuracy on the most complex business queries. --- ### Learning PDE Solvers with Physics and Data: A Unifying View of Physics-Informed Neural Networks and Neural Operators Authors: Yilong Dai, Shengyu Chen, Ziyi Wang, Xiaowei Jia, Yiqun Xie, Vipin Kumar, Runlong Yu Published: 2026-01-20 Categories: cs.LG, math.AP arXiv: https://arxiv.org/abs/2601.14517v2 Partial differential equations (PDEs) are central to scientific modeling. Modern workflows increasingly rely on learning-based components to support model reuse, inference, and integration across large computational processes. Despite the emergence of various physics-aware data-driven approaches, the field still lacks a unified perspective to uncover their relationships, limitations, and appropriate roles in scientific workflows. To this end, we propose a unifying perspective to place two dominant paradigms: Physics-Informed Neural Networks (PINNs) and Neural Operators (NOs), within a shared design space. We organize existing methods from three fundamental dimensions: what is learned, how physical structures are integrated into the learning process, and how the computational load is amortized across problem instances. In this way, many challenges can be best understood as consequences of these structural properties of learning PDEs. By analyzing advances through this unifying view, our survey aims to facilitate the development of reliable learning-based PDE solvers and catalyze a synthesis of physics and data. ---