# Research Papers — 2026-02 · bundle page 104 of 210

> Full-text markdown of papers #5151–5200 of 10493 published in 2026-02, newest first. AI-readable mirrors of the full 10493-paper 2026-02 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-02/page/104
[← Previous bundle](https://aipod.dev/papers/archive/2026-02/page/103.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-02/page/105.md)

### Geometry of Uncertainty: Learning Metric Spaces for Multimodal State Estimation in RL

Authors: Alfredo Reichlin, Adriano Pacciarelli, Danica Kragic, Miguel Vasco
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.12087v1

Estimating the state of an environment from high-dimensional, multimodal, and noisy observations is a fundamental challenge in reinforcement learning (RL). Traditional approaches rely on probabilistic models to account for the uncertainty, but often require explicit noise assumptions, in turn limiting generalization. In this work, we contribute a novel method to learn a structured latent representation, in which distances between states directly correlate with the minimum number of actions required to transition between them. The proposed metric space formulation provides a geometric interpretation of uncertainty without the need for explicit probabilistic modeling. To achieve this, we introduce a multimodal latent transition model and a sensor fusion mechanism based on inverse distance weighting, allowing for the adaptive integration of multiple sensor modalities without prior knowledge of noise distributions. We empirically validate the approach on a range of multimodal RL tasks, demonstrating improved robustness to sensor noise and superior state estimation compared to baseline methods. Our experiments show enhanced performance of an RL agent via the learned representation, eliminating the need of explicit noise augmentation. The presented results suggest that leveraging transition-aware metric spaces provides a principled and scalable solution for robust state estimation in sequential decision-making.

---

### Differentiable Modal Logic for Multi-Agent Diagnosis, Orchestration and Communication

Authors: Antonin Sulc
Published: 2026-02-12
Categories: cs.AI, cs.LO
arXiv: https://arxiv.org/abs/2602.12083v1

As multi-agent AI systems evolve from simple chatbots to autonomous swarms, debugging semantic failures requires reasoning about knowledge, belief, causality, and obligation, precisely what modal logic was designed to formalize. However, traditional modal logic requires manual specification of relationship structures that are unknown or dynamic in real systems. This tutorial demonstrates differentiable modal logic (DML), implemented via Modal Logical Neural Networks (MLNNs), enabling systems to learn trust networks, causal chains, and regulatory boundaries from behavioral data alone. We present a unified neurosymbolic debugging framework through four modalities: epistemic (who to trust), temporal (when events cause failures), deontic (what actions are permitted), and doxastic (how to interpret agent confidence). Each modality is demonstrated on concrete multi-agent scenarios, from discovering deceptive alliances in diplomacy games to detecting LLM hallucinations, with complete implementations showing how logical contradictions become learnable optimization objectives. Key contributions for the neurosymbolic community: (1) interpretable learned structures where trust and causality are explicit parameters, not opaque embeddings; (2) knowledge injection via differentiable axioms that guide learning with sparse data (3) compositional multi-modal reasoning that combines epistemic, temporal, and deontic constraints; and (4) practical deployment patterns for monitoring, active control and communication of multi-agent systems. All code provided as executable Jupyter notebooks.

---

### Empirical Gaussian Processes

Authors: Jihao Andreas Lin, Sebastian Ament, Louis C. Tiao, David Eriksson, Maximilian Balandat, Eytan Bakshy
Published: 2026-02-12
Categories: cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2602.12082v1

Gaussian processes (GPs) are powerful and widely used probabilistic regression models, but their effectiveness in practice is often limited by the choice of kernel function. This kernel function is typically handcrafted from a small set of standard functions, a process that requires expert knowledge, results in limited adaptivity to data, and imposes strong assumptions on the hypothesis space. We study Empirical GPs, a principled framework for constructing flexible, data-driven GP priors that overcome these limitations. Rather than relying on standard parametric kernels, we estimate the mean and covariance functions empirically from a corpus of historical observations, enabling the prior to reflect rich, non-trivial covariance structures present in the data. Theoretically, we show that the resulting model converges to the GP that is closest (in KL-divergence sense) to the real data generating process. Practically, we formulate the problem of learning the GP prior from independent datasets as likelihood estimation and derive an Expectation-Maximization algorithm with closed-form updates, allowing the model handle heterogeneous observation locations across datasets. We demonstrate that Empirical GPs achieve competitive performance on learning curve extrapolation and time series forecasting benchmarks.

---

### PathCRF: Ball-Free Soccer Event Detection via Possession Path Inference from Player Trajectories

Authors: Hyunsung Kim, Kunhee Lee, Sangwoo Seo, Sang-Ki Ko, Jinsung Yoon, Chanyoung Park
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.12080v2

Despite recent advances in AI, event data collection in soccer still relies heavily on labor-intensive manual annotation. Although prior work has explored automatic event detection using player and ball trajectories, ball tracking also remains difficult to scale due to high infrastructural and operational costs. As a result, comprehensive data collection in soccer is largely confined to top-tier competitions, limiting the broader adoption of data-driven analysis in this domain. To address this challenge, this paper proposes PathCRF, a framework for detecting on-ball soccer events using only player tracking data. We model player trajectories as a fully connected dynamic graph and formulate event detection as the problem of selecting exactly one edge corresponding to the current possession state at each time step. To ensure logical consistency of the resulting edge sequence, we employ a Conditional Random Field (CRF) that forbids impossible transitions between consecutive edges, where emission and transition scores are dynamically computed from edge embeddings produced by a socio-temporal backbone architecture. During inference, the most probable edge sequence is obtained via Viterbi decoding, and events such as ball controls or passes are detected whenever the selected edge changes between adjacent time steps. Experiments show that PathCRF produces accurate, logically consistent possession paths, enabling reliable downstream analyses while substantially reducing the need for manual event annotation. The source code is available at https://github.com/hyunsungkim-ds/pathcrf.git.

---

### Tiny Recursive Reasoning with Mamba-2 Attention Hybrid

Authors: Wenlong Wang, Fergal Reid
Published: 2026-02-12
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2602.12078v2

Recent work on recursive reasoning models like TRM demonstrates that tiny networks (7M parameters) can achieve strong performance on abstract reasoning tasks through latent recursion -- iterative refinement in hidden representation space without emitting intermediate tokens. This raises a natural question about operator choice: Mamba-2's state space recurrence is itself a form of iterative refinement, making it a natural candidate for recursive reasoning -- but does introducing Mamba-2 into the recursive scaffold preserve reasoning capability? We investigate this by replacing the Transformer blocks in TRM with Mamba-2 hybrid operators while maintaining parameter parity (6.83M vs 6.86M parameters). On ARC-AGI-1, we find that the hybrid improves pass@2 (the official metric) by +2.0\% (45.88\% vs 43.88\%) and consistently outperforms at higher K values (+4.75\% at pass@100), whilst maintaining pass@1 parity. This suggests improved candidate coverage -- the model generates correct solutions more reliably -- with similar top-1 selection. Our results validate that Mamba-2 hybrid operators preserve reasoning capability within the recursive scaffold, establishing SSM-based operators as viable candidates in the recursive operator design space and taking a first step towards understanding the best mixing strategies for recursive reasoning.

---

### ModelWisdom: An Integrated Toolkit for TLA+ Model Visualization, Digest and Repair

Authors: Zhiyong Chen, Jialun Cao, Chang Xu, Shing-Chi Cheung
Published: 2026-02-12
Categories: cs.SE, cs.AI, cs.FL
arXiv: https://arxiv.org/abs/2602.12058v1

Model checking in TLA+ provides strong correctness guarantees, yet practitioners continue to face significant challenges in interpreting counterexamples, understanding large state-transition graphs, and repairing faulty models. These difficulties stem from the limited explainability of raw model-checker output and the substantial manual effort required to trace violations back to source specifications. Although the TLA+ Toolbox includes a state diagram viewer, it offers only a static, fully expanded graph without folding, color highlighting, or semantic explanations, which limits its scalability and interpretability. We present ModelWisdom, an interactive environment that uses visualization and large language models to make TLA+ model checking more interpretable and actionable. ModelWisdom offers: (i) Model Visualization, with colorized violation highlighting, click-through links from transitions to TLA+ code, and mapping between violating states and broken properties; (ii) Graph Optimization, including tree-based structuring and node/edge folding to manage large models; (iii) Model Digest, which summarizes and explains subgraphs via large language models (LLMs) and performs preprocessing and partial explanations; and (iv) Model Repair, which extracts error information and supports iterative debugging. Together, these capabilities turn raw model-checker output into an interactive, explainable workflow, improving understanding and reducing debugging effort for nontrivial TLA+ specifications. The website to ModelWisdom is available: https://model-wisdom.pages.dev. A demonstrative video can be found at https://www.youtube.com/watch?v=plyZo30VShA.

---

### LawThinker: A Deep Research Legal Agent in Dynamic Environments

Authors: Xinyu Yang, Chenlong Deng, Tongyu Wen, Binyu Xie, Zhicheng Dou
Published: 2026-02-12
Categories: cs.AI
arXiv: https://arxiv.org/abs/2602.12056v1

Legal reasoning requires not only correct outcomes but also procedurally compliant reasoning processes. However, existing methods lack mechanisms to verify intermediate reasoning steps, allowing errors such as inapplicable statute citations to propagate undetected through the reasoning chain. To address this, we propose LawThinker, an autonomous legal research agent that adopts an Explore-Verify-Memorize strategy for dynamic judicial environments. The core idea is to enforce verification as an atomic operation after every knowledge exploration step. A DeepVerifier module examines each retrieval result along three dimensions of knowledge accuracy, fact-law relevance, and procedural compliance, with a memory module for cross-round knowledge reuse in long-horizon tasks. Experiments on the dynamic benchmark J1-EVAL show that LawThinker achieves a 24% improvement over direct reasoning and an 11% gain over workflow-based methods, with particularly strong improvements on process-oriented metrics. Evaluations on three static benchmarks further confirm its generalization capability. The code is available at https://github.com/yxy-919/LawThinker-agent .

---

### Multi UAVs Preflight Planning in a Shared and Dynamic Airspace

Authors: Amath Sow, Mauricio Rodriguez Cesen, Fabiola Martins Campos de Oliveira, Mariusz Wzorek, Daniel de Leng, Mattias Tiger, Fredrik Heintz, Christian Esteve Rothenberg
Published: 2026-02-12
Categories: cs.AI, cs.MA, cs.RO
arXiv: https://arxiv.org/abs/2602.12055v1

Preflight planning for large-scale Unmanned Aerial Vehicle (UAV) fleets in dynamic, shared airspace presents significant challenges, including temporal No-Fly Zones (NFZs), heterogeneous vehicle profiles, and strict delivery deadlines. While Multi-Agent Path Finding (MAPF) provides a formal framework, existing methods often lack the scalability and flexibility required for real-world Unmanned Traffic Management (UTM). We propose DTAPP-IICR: a Delivery-Time Aware Prioritized Planning method with Incremental and Iterative Conflict Resolution. Our framework first generates an initial solution by prioritizing missions based on urgency. Secondly, it computes roundtrip trajectories using SFIPP-ST, a novel 4D single-agent planner (Safe Flight Interval Path Planning with Soft and Temporal Constraints). SFIPP-ST handles heterogeneous UAVs, strictly enforces temporal NFZs, and models inter-agent conflicts as soft constraints. Subsequently, an iterative Large Neighborhood Search, guided by a geometric conflict graph, efficiently resolves any residual conflicts. A completeness-preserving directional pruning technique further accelerates the 3D search. On benchmarks with temporal NFZs, DTAPP-IICR achieves near-100% success with fleets of up to 1,000 UAVs and gains up to 50% runtime reduction from pruning, outperforming batch Enhanced Conflict-Based Search in the UTM context. Scaling successfully in realistic city-scale operations where other priority-based methods fail even at moderate deployments, DTAPP-IICR is positioned as a practical and scalable solution for preflight planning in dense, dynamic urban airspace.

---

### Improving HPC Code Generation Capability of LLMs via Online Reinforcement Learning with Real-Machine Benchmark Rewards

Authors: Ryo Mikasa, Shun-ichiro Hayashi, Daichi Mukunoki, Tetsuya Hoshino, Takahiro Katagiri
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.12049v1

Large language models (LLMs) have demonstrated strong code generation capabilities, yet the runtime performance of generated code is not guaranteed, and there have been few attempts to train LLMs using runtime performance as a reward in the HPC domain. We propose an online reinforcement learning approach that executes LLM-generated code on a supercomputer and directly feeds back the measured runtime performance (GFLOPS) as a reward. We further introduce a Staged Quality-Diversity (SQD) algorithm that progressively varies the permitted optimization techniques on a per-problem basis, enabling the model to learn code optimization from diverse perspectives. We build a distributed system connecting a GPU training cluster with a CPU benchmarking cluster, and train Qwen2.5 Coder 14B on a double-precision matrix multiplication task using Group Relative Policy Optimization (GRPO). Through two experiments, we show that reinforcement learning combining runtime performance feedback with staged optimization can improve the HPC code generation capability of LLMs.

---

### Safety Beyond the Training Data: Robust Out-of-Distribution MPC via Conformalized System Level Synthesis

Authors: Anutam Srinivasan, Antoine Leeman, Glen Chou
Published: 2026-02-12
Categories: cs.RO, cs.LG, eess.SY, math.OC
arXiv: https://arxiv.org/abs/2602.12047v1

We present a novel framework for robust out-of-distribution planning and control using conformal prediction (CP) and system level synthesis (SLS), addressing the challenge of ensuring safety and robustness when using learned dynamics models beyond the training data distribution. We first derive high-confidence model error bounds using weighted CP with a learned, state-control-dependent covariance model. These bounds are integrated into an SLS-based robust nonlinear model predictive control (MPC) formulation, which performs constraint tightening over the prediction horizon via volume-optimized forward reachable sets. We provide theoretical guarantees on coverage and robustness under distributional drift, and analyze the impact of data density and trajectory tube size on prediction coverage. Empirically, we demonstrate our method on nonlinear systems of increasing complexity, including a 4D car and a {12D} quadcopter, improving safety and robustness compared to fixed-bound and non-robust baselines, especially outside of the data distribution.

---

### Fourier Transformers for Latent Crystallographic Diffusion and Generative Modeling

Authors: Jed A. Duersch, Elohan Veillon, Astrid Klipfel, Adlane Sayede, Zied Bouraoui
Published: 2026-02-12
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.12045v1

The discovery of new crystalline materials calls for generative models that handle periodic boundary conditions, crystallographic symmetries, and physical constraints, while scaling to large and structurally diverse unit cells. We propose a reciprocal-space generative pipeline that represents crystals through a truncated Fourier transform of the species-resolved unit-cell density, rather than modeling atomic coordinates directly. This representation is periodicity-native, admits simple algebraic actions of space-group symmetries, and naturally supports variable atomic multiplicities during generation, addressing a common limitation of particle-based approaches. Using only nine Fourier basis functions per spatial dimension, our approach reconstructs unit cells containing up to 108 atoms per chemical species. We instantiate this pipeline with a transformer variational autoencoder over complex-valued Fourier coefficients, and a latent diffusion model that generates in the compressed latent space. We evaluate reconstruction and latent diffusion on the LeMaterial benchmark and compare unconditional generation against coordinate-based baselines in the small-cell regime ($\leq 16$ atoms per unit cell).

---

### A DMD-Based Adaptive Modulation Method for High Dynamic Range Imaging in High-Glare Environments

Authors: Banglei Guan, Jing Tao, Liang Xu, Dongcai Tan, Pengju Sun, Jianbing Liu, Yang Shang, Qifeng Yu
Published: 2026-02-12
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.12044v2

Background The accuracy of photomechanics measurements critically relies on image quality,particularly under extreme illumination conditions such as welding arc monitoring and polished metallic surface analysis. High dynamic range (HDR) imaging above 120 dB is essential in these contexts. Conventional CCD/CMOS sensors, with dynamic ranges typically below 70 dB, are highly susceptible to saturation under glare, resulting in irreversible loss of detail and significant errors in digital image correlation (DIC). Methods This paper presents an HDR imaging system that leverages the spatial modulation capability of a digital micromirror device (DMD). The system architecture enables autonomous regional segmentation and adaptive exposure control for high-dynamic-range scenes through an integrated framework comprising two synergistic subsystems: a DMD-based optical modulation unit and an adaptive computational imaging pipeline. Results The system achieves a measurable dynamic range of 127 dB, effectively eliminating satu ration artifacts under high glare. Experimental results demonstrate a 78% reduction in strain error and improved DIC positioning accuracy, confirming reliable performance across extreme intensity variations. Conclusion The DMD-based system provides high fidelity adaptive HDR imaging, overcoming key limitations of conventional sensors. It exhibits strong potential for optical metrology and stress analysis in high-glare environments where traditional methods are inadequate.

---

### Improved Inference for CSDID Using the Cluster Jackknife

Authors: Sunny R. Karim, Morten Ørregaard Nielsen, James G. MacKinnon, Matthew D. Webb
Published: 2026-02-12
Categories: econ.EM, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2602.12043v1

Obtaining reliable inferences with traditional difference-in-differences (DiD) methods can be difficult. Problems can arise when both outcomes and errors are serially correlated, when there are few clusters or few treated clusters, when cluster sizes vary greatly, and in various other cases. In recent years, recognition of the ``staggered adoption'' problem has shifted the focus away from inference towards consistent estimation of treatment effects. One of the most popular new estimators is the CSDID procedure of Callaway and Sant'Anna (2021). We find that the issues of over-rejection with few clusters and/or few treated clusters are at least as severe for CSDID as for traditional DiD methods. We also propose using a cluster jackknife for inference with CSDID, which simulations suggest greatly improves inference. We provide software packages in Stata csdidjack and R didjack to calculate cluster-jackknife standard errors easily.

---

### The Implicit Bias of Logit Regularization

Authors: Alon Beck, Yohai Bar Sinai, Noam Levi
Published: 2026-02-12
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2602.12039v2

Logit regularization, the addition of a convex penalty directly in logit space, is widely used in modern classifiers, with label smoothing as a prominent example. While such methods often improve calibration and generalization, their mechanism remains under-explored. In this work, we analyze a general class of such logit regularizers in the context of linear classification, and demonstrate that they induce an implicit bias of logit clustering around finite per-sample targets. For Gaussian data, or whenever logits are sufficiently clustered, we prove that logit clustering drives the weight vector to align exactly with Fisher's Linear Discriminant. To demonstrate the consequences, we study a simple signal-plus-noise model in which this transition has dramatic effects: Logit regularization halves the critical sample complexity and induces grokking in the small-noise limit, while making generalization robust to noise. Our results extend the theoretical understanding of label smoothing and highlight the efficacy of a broader class of logit-regularization methods.

---

### An Empirical Study of the Imbalance Issue in Software Vulnerability Detection

Authors: Yuejun Guo, Qiang Hu, Qiang Tang, Yves Le Traon
Published: 2026-02-12
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2602.12038v1

Vulnerability detection is crucial to protect software security. Nowadays, deep learning (DL) is the most promising technique to automate this detection task, leveraging its superior ability to extract patterns and representations within extensive code volumes. Despite its promise, DL-based vulnerability detection remains in its early stages, with model performance exhibiting variability across datasets. Drawing insights from other well-explored application areas like computer vision, we conjecture that the imbalance issue (the number of vulnerable code is extremely small) is at the core of the phenomenon. To validate this, we conduct a comprehensive empirical study involving nine open-source datasets and two state-of-the-art DL models. The results confirm our conjecture. We also obtain insightful findings on how existing imbalance solutions perform in vulnerability detection. It turns out that these solutions perform differently as well across datasets and evaluation metrics. Specifically: 1) Focal loss is more suitable to improve the precision, 2) mean false error and class-balanced loss encourages the recall, and 3) random over-sampling facilitates the F1-measure. However, none of them excels across all metrics. To delve deeper, we explore external influences on these solutions and offer insights for developing new solutions.

---

### Composition-RL: Compose Your Verifiable Prompts for Reinforcement Learning of Large Language Models

Authors: Xin Xu, Clive Bai, Kai Yang, Tianhao Chen, Yangkun Chen, Weijie Liu, Hao Chen, Yang Wang, Saiyong Yang, Can Yang
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.12036v2

Large-scale verifiable prompts underpin the success of Reinforcement Learning with Verifiable Rewards (RLVR), but they contain many uninformative examples and are costly to expand further. Recent studies focus on better exploiting limited training data by prioritizing hard prompts whose rollout pass rate is 0. However, easy prompts with a pass rate of 1 also become increasingly prevalent as training progresses, thereby reducing the effective data size. To mitigate this, we propose Composition-RL, a simple yet useful approach for better utilizing limited verifiable prompts targeting pass-rate-1 prompts. More specifically, Composition-RL automatically composes multiple problems into a new verifiable question and uses these compositional prompts for RL training. Extensive experiments across model sizes from 4B to 30B show that Composition-RL consistently improves reasoning capability over RL trained on the original dataset. Performance can be further boosted with a curriculum variant of Composition-RL that gradually increases compositional depth over training. Additionally, Composition-RL enables more effective cross-domain RL by composing prompts drawn from different domains. Codes, datasets, and models are available at https://github.com/XinXU-USTC/Composition-RL.

---

### PrefillShare: A Shared Prefill Module for KV Reuse in Multi-LLM Disaggregated Serving

Authors: Sunghyeon Woo, Hoseung Kim, Sunghwan Shim, Minjung Jo, Hyunjoon Jeong, Jeongtae Lee, Joonghoon Kim, Sungjae Lee, Jiwon Ryu, Baeseong Park, Se Jung Kwon, Dongsoo Lee
Published: 2026-02-12
Categories: cs.LG, cs.DC
arXiv: https://arxiv.org/abs/2602.12029v2

Multi-agent systems increasingly orchestrate multiple specialized language models to solve complex real-world problems, often invoking them over a shared context. This execution pattern repeatedly processes the same prompt prefix across models. Consequently, each model redundantly executes the prefill stage and maintains its own key-value (KV) cache, increasing aggregate prefill load and worsening tail latency by intensifying prefill-decode interference in existing LLM serving stacks. Disaggregated serving reduces such interference by placing prefill and decode on separate GPUs, but disaggregation does not fundamentally eliminate inter-model redundancy in computation and KV storage for the same prompt. To address this issue, we propose PrefillShare, a novel algorithm that enables sharing the prefill stage across multiple fine-tuned models in a disaggregated setting. PrefillShare factorizes the model into prefill and decode modules, freezes the prefill module, and fine-tunes only the decode module. This design allows multiple task-specific models to share a prefill module and the KV cache generated for the same prompt. We further introduce a routing mechanism that enables effective prefill sharing in a vLLM-based disaggregated system. PrefillShare not only matches full fine-tuning accuracy on a broad range of tasks and models, but also delivers 4.5x lower p95 latency and 2.0x higher throughput in multi-model agent workloads.

---

### PrefillShare: A Shared Prefill Module for KV Reuse in Multi-LLM Disaggregated Serving

Authors: Sunghyeon Woo, Hoseung Kim, Sunghwan Shim, Minjung Jo, Hyunjoon Jeong, Jeongtae Lee, Joonghoon Kim, Sungjae Lee, Baeseong Park, Se Jung Kwon, Dongsoo Lee
Published: 2026-02-12
Categories: cs.LG, cs.DC
arXiv: https://arxiv.org/abs/2602.12029v1

Multi-agent systems increasingly orchestrate multiple specialized language models to solve complex real-world problems, often invoking them over a shared context. This execution pattern repeatedly processes the same prompt prefix across models. Consequently, each model redundantly executes the prefill stage and maintains its own key-value (KV) cache, increasing aggregate prefill load and worsening tail latency by intensifying prefill-decode interference in existing LLM serving stacks. Disaggregated serving reduces such interference by placing prefill and decode on separate GPUs, but disaggregation does not fundamentally eliminate inter-model redundancy in computation and KV storage for the same prompt. To address this issue, we propose PrefillShare, a novel algorithm that enables sharing the prefill stage across multiple models in a disaggregated setting. PrefillShare factorizes the model into prefill and decode modules, freezes the prefill module, and fine-tunes only the decode module. This design allows multiple task-specific models to share a prefill module and the KV cache generated for the same prompt. We further introduce a routing mechanism that enables effective prefill sharing across heterogeneous models in a vLLM-based disaggregated system. PrefillShare not only matches full fine-tuning accuracy on a broad range of tasks and models, but also delivers 4.5x lower p95 latency and 3.9x higher throughput in multi-model agent workloads.

---

### Protein Circuit Tracing via Cross-layer Transcoders

Authors: Darin Tsui, Kunal Talreja, Daniel Saeedi, Amirali Aghazadeh
Published: 2026-02-12
Categories: cs.LG, q-bio.QM
arXiv: https://arxiv.org/abs/2602.12026v2

Protein language models (pLMs) have emerged as powerful predictors of protein structure and function. However, the computational circuits underlying their predictions remain poorly understood. Recent mechanistic interpretability methods decompose pLM representations into interpretable features, but they treat each layer independently and thus fail to capture cross-layer computation, limiting their ability to approximate the full model. We introduce ProtoMech, a framework for discovering computational circuits in pLMs using cross-layer transcoders that learn sparse latent representations jointly across layers to capture the model's full computational circuitry. Applied to the pLM ESM2, ProtoMech recovers 82-89% of the original performance on protein family classification and function prediction tasks. ProtoMech then identifies compressed circuits that use <1% of the latent space while retaining up to 79% of model accuracy, revealing correspondence with structural and functional motifs, including binding, signaling, and stability. Steering along these circuits enables high-fitness protein design, surpassing baseline methods in more than 70% of cases. These results establish ProtoMech as a principled framework for protein circuit tracing.

---

### Decomposition of Spillover Effects Under Misspecification: Pseudo-true Estimands and a Local-Global Extension

Authors: Yechan Park, Xiaodong Yang
Published: 2026-02-12
Categories: econ.EM, math.ST, stat.ML
arXiv: https://arxiv.org/abs/2602.12023v4

To measure spillovers, researchers often summarize who else was treated with a simple measure, such as the share of treated neighbors. We study what the researcher estimates when that summary is misspecified. We show that the researcher estimates the best approximation to the true policy effect among all functionals of the chosen summary, and that the usual direct and spillover estimates are exactly the components of this approximation. Under a monotonicity restriction, the estimates also preserve the signs of the true effects. We then specialize this framework to a setting common in applications such as cash transfers: treatment spills over both globally, through market equilibrium, and locally, through network externalities, while the researcher models only one channel. The network estimator still recovers the network spillover, and the equilibrium estimator the equilibrium spillover, even when the other channel is ignored. We illustrate with a simulation calibrated to a large cash-transfer experiment.

---

### Improved state mixing in higher-order and block diagonal linear recurrent networks

Authors: Igor Dubinin, Antonio Orvieto, Felix Effenberger
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.12021v2

Linear recurrent networks (LRNNs) and linear state space models (SSMs) promise computational and memory efficiency on long-sequence modeling tasks, yet their diagonal state transitions limit expressivity. Dense and nonlinear architectures (e.g., LSTMs) on the other hand are provably more expressive, but computationally costly. Here, we explore how expressivity in LRNNs can be increased via richer state mixing across time and channels while maintaining competitive efficiency. Specifically, we introduce two structured LRNN architectures: (i) Higher-order Linear Recurrent Units (H-LRU), which generalize first-order recurrence to higher order, mixing multiple past states, and (ii) Block-Diagonal LRUs (BD-LRU), which enable dense intra-block channel mixing. Per-channel (H-LRU) or per-row (BD-LRU) L1-normalization of selective gates stabilizes training and allows for scaling window/block sizes. A parallel-scan implementation of the proposed architectures keeps the throughput competitive with diagonal LRNNs for moderate orders (H-LRU) and block sizes (BD-LRU). In synthetic sequence modeling tasks, the performance of BD-LRU matches or exceeds those of linear SSMs (Mamba), low-rank LRNNs (DeltaNet) and LSTM baselines, while H-LRU is found to be the most parameter-efficient in compression task. In both synthetic sequence modeling and language modeling, our results indicate that the structure of state mixing rather than width alone shapes expressivity of LRNNs, offering a practical route to closing the efficiency-expressivity gap in linear sequence models.

---

### Artificial intelligence is creating a new global linguistic hierarchy

Authors: Giulia Occhini, Kumiko Tanaka-Ishii, Anna Barford, Refael Tikochinski, Songbo Hu, Roi Reichart, Yijie Zhou, Hannah Claus, Ulla Petti, Ivan Vulić, Ramit Debnath, Anna Korhonen
Published: 2026-02-12
Categories: cs.CY, cs.CL
arXiv: https://arxiv.org/abs/2602.12018v1

Artificial intelligence (AI) has the potential to transform healthcare, education, governance and socioeconomic equity, but its benefits remain concentrated in a small number of languages (Bender, 2019; Blasi et al., 2022; Joshi et al., 2020; Ranathunga and de Silva, 2022; Young, 2015). Language AI - the technologies that underpin widely-used conversational systems such as ChatGPT - could provide major benefits if available in people's native languages, yet most of the world's 7,000+ linguistic communities currently lack access and face persistent digital marginalization. Here we present a global longitudinal analysis of social, economic and infrastructural conditions across languages to assess systemic inequalities in language AI. We first analyze the existence of AI resources for 6003 languages. We find that despite efforts of the community to broaden the reach of language technologies (Bapna et al., 2022; Costa-Jussà et al., 2022), the dominance of a handful of languages is exacerbating disparities on an unprecedented scale, with divides widening exponentially rather than narrowing. Further, we contrast the longitudinal diffusion of AI with that of earlier IT technologies, revealing a distinctive hype-driven pattern of spread. To translate our findings into practical insights and guide prioritization efforts, we introduce the Language AI Readiness Index (EQUATE), which maps the state of technological, socio-economic, and infrastructural prerequisites for AI deployment across languages. The index highlights communities where capacity exists but remains underutilized, and provides a framework for accelerating more equitable diffusion of language AI. Our work contributes to setting the baseline for a transition towards more sustainable and equitable language technologies.

---

### Disentangling Ambiguity from Instability in Large Language Models: A Clinical Text-to-SQL Case Study

Authors: Angelo Ziletti, Leonardo D'Ambrosi
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.12015v2

Deploying large language models for clinical Text-to-SQL requires distinguishing two qualitatively different causes of output diversity: (i) input ambiguity that should trigger clarification, and (ii) model instability that should trigger human review. We propose CLUES, a framework that models Text-to-SQL as a two-stage process (interpretations --> answers) and decomposes semantic uncertainty into an ambiguity score and an instability score. The instability score is computed via the Schur complement of a bipartite semantic graph matrix. Across AmbigQA/SituatedQA (gold interpretations) and a clinical Text-to-SQL benchmark (known interpretations), CLUES improves failure prediction over state-of-the-art Kernel Language Entropy. In deployment settings, it remains competitive while providing a diagnostic decomposition unavailable from a single score. The resulting uncertainty regimes map to targeted interventions - query refinement for ambiguity, model improvement for instability. The high-ambiguity/high-instability regime contains 51% of errors while covering 25% of queries, enabling efficient triage.

---

### FedGRPO: Privately Optimizing Foundation Models with Group-Relative Rewards from Domain Client

Authors: Gongxi Zhu, Hanlin Gu, Lixin Fan, Qiang Yang, Yuxing Han
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.12014v1

One important direction of Federated Foundation Models (FedFMs) is leveraging data from small client models to enhance the performance of a large server-side foundation model. Existing methods based on model level or representation level knowledge transfer either require expensive local training or incur high communication costs and introduce unavoidable privacy risks. We reformulate this problem as a reinforcement learning style evaluation process and propose FedGRPO, a privacy preserving framework comprising two modules. The first module performs competence-based expert selection by building a lightweight confidence graph from auxiliary data to identify the most suitable clients for each question. The second module leverages the "Group Relative" concept from the Group Relative Policy Optimization (GRPO) framework by packaging each question together with its solution rationale into candidate policies, dispatching these policies to a selected subset of expert clients, and aggregating solely the resulting scalar reward signals via a federated group-relative loss function. By exchanging reward values instead of data or model updates, FedGRPO reduces privacy risk and communication overhead while enabling parallel evaluation across heterogeneous devices. Empirical results on diverse domain tasks demonstrate that FedGRPO achieves superior downstream accuracy and communication efficiency compared to conventional FedFMs baselines.

---

### InjectRBP: Steering Large Language Model Reasoning Behavior via Pattern Injection

Authors: Xiuping Wu, Zhao Yu, Yuxin Cheng, Ngai Wong, Liangjun Ke, Tapas Mishra, Konstantinos V. Katsikopoulos
Published: 2026-02-12
Categories: cs.AI
arXiv: https://arxiv.org/abs/2602.12013v1

Reasoning can significantly enhance the performance of Large Language Models. While recent studies have exploited behavior-related prompts adjustment to enhance reasoning, these designs remain largely intuitive and lack a systematic analysis of the underlying behavioral patterns. Motivated by this, we investigate how models' reasoning behaviors shape reasoning from the perspective of behavioral patterns. We observe that models exhibit adaptive distributions of reasoning behaviors when responding to specific types of questions, and that structurally injecting these patterns can substantially influence the quality of the models' reasoning processes and outcomes. Building on these findings, we propose two optimization methods that require no parameter updates: InjectCorrect and InjectRLOpt. InjectCorrect guides the model by imitating behavioral patterns derived from its own past correct answers. InjectRLOpt learns a value function from historical behavior-pattern data and, via our proposed Reliability-Aware Softmax Policy, generates behavioral injectant during inference to steer the reasoning process. Our experiments demonstrate that both methods can improve model performance across various reasoning tasks without requiring any modifications to model parameters, achieving gains of up to 5.34% and 8.67%, respectively.

---

### On the Sensitivity of Firing Rate-Based Federated Spiking Neural Networks to Differential Privacy

Authors: Luiz Pereira, Mirko Perkusich, Dalton Valadares, Kyller Gorgônio
Published: 2026-02-12
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.12009v1

Federated Neuromorphic Learning (FNL) enables energy-efficient and privacy-preserving learning on devices without centralizing data. However, real-world deployments require additional privacy mechanisms that can significantly alter training signals. This paper analyzes how Differential Privacy (DP) mechanisms, specifically gradient clipping and noise injection, perturb firing-rate statistics in Spiking Neural Networks (SNNs) and how these perturbations are propagated to rate-based FNL coordination. On a speech recognition task under non-IID settings, ablations across privacy budgets and clipping bounds reveal systematic rate shifts, attenuated aggregation, and ranking instability during client selection. Moreover, we relate these shifts to sparsity and memory indicators. Our findings provide actionable guidance for privacy-preserving FNL, specifically regarding the balance between privacy strength and rate-dependent coordination.

---

### LaCy: What Small Language Models Can and Should Learn is Not Just a Question of Loss

Authors: Szilvia Ujváry, Louis Béthune, Pierre Ablin, João Monteiro, Marco Cuturi, Michael Kirchhof
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.12005v4

Language models have consistently grown to compress more world knowledge into their parameters, but the knowledge that can be pretrained into them is upper-bounded by their parameter size. Especially the capacity of Small Language Models (SLMs) is limited, leading to factually incorrect generations. This problem is often mitigated by giving the SLM access to an outside source: the ability to query a larger model, documents, or a database. Under this setting, we study the fundamental question of \emph{which tokens an SLM can and should learn} during pretraining, versus \emph{which ones it should delegate} via a \texttt{<CALL>} token. We find that this is not simply a question of loss: although the loss is predictive of whether a predicted token mismatches the ground-truth, it is insufficient for identifying which predictions would actually lead to factual or semantically invalid continuations. Some high-loss tokens correspond to \emph{acceptable} alternative continuations of a pretraining document and therefore should not trigger a \texttt{<CALL>}. This suggests that learnability cannot be characterized from loss alone, but requires additional domain-specific signals about the role of a token in the sentence. In Wikipedia-like domains, we show that augmenting the loss signal with lightweight grammatical information from a spaCy parser substantially improves delegation decisions. Based on this insight, we propose LaCy, a novel pretraining method that combines loss with factuality signals to decide which tokens an SLM should learn. Our experiments demonstrate that LaCy models successfully learn which tokens to predict and when to call for help. This results in higher FactScores when generating in a cascade with a bigger model and outperforms Rho or LLM-judge trained SLMs, while being simpler and cheaper.

---

### CSEval: A Framework for Evaluating Clinical Semantics in Text-to-Image Generation

Authors: Robert Cronshaw, Konstantinos Vilouras, Junyu Yan, Yuning Du, Feng Chen, Steven McDonagh, Sotirios A. Tsaftaris
Published: 2026-02-12
Categories: cs.AI
arXiv: https://arxiv.org/abs/2602.12004v1

Text-to-image generation has been increasingly applied in medical domains for various purposes such as data augmentation and education. Evaluating the quality and clinical reliability of these generated images is essential. However, existing methods mainly assess image realism or diversity, while failing to capture whether the generated images reflect the intended clinical semantics, such as anatomical location and pathology. In this study, we propose the Clinical Semantics Evaluator (CSEval), a framework that leverages language models to assess clinical semantic alignment between the generated images and their conditioning prompts. Our experiments show that CSEval identifies semantic inconsistencies overlooked by other metrics and correlates with expert judgment. CSEval provides a scalable and clinically meaningful complement to existing evaluation methods, supporting the safe adoption of generative models in healthcare.

---

### Projected Representation Conditioning for High-fidelity Novel View Synthesis

Authors: Min-Seop Kwak, Minkyung Kwon, Jinhyeok Choi, Jiho Park, Seungryong Kim
Published: 2026-02-12
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.12003v1

We propose a novel framework for diffusion-based novel view synthesis in which we leverage external representations as conditions, harnessing their geometric and semantic correspondence properties for enhanced geometric consistency in generated novel viewpoints. First, we provide a detailed analysis exploring the correspondence capabilities emergent in the spatial attention of external visual representations. Building from these insights, we propose a representation-guided novel view synthesis through dedicated representation projection modules that inject external representations into the diffusion process, a methodology named ReNoV, short for representation-guided novel view synthesis. Our experiments show that this design yields marked improvements in both reconstruction fidelity and inpainting quality, outperforming prior diffusion-based novel-view methods on standard benchmarks and enabling robust synthesis from sparse, unposed image collections.

---

### Can Local Vision-Language Models improve Activity Recognition over Vision Transformers? -- Case Study on Newborn Resuscitation

Authors: Enrico Guerriero, Kjersti Engan, Øyvind Meinich-Bache
Published: 2026-02-12
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.12002v1

Accurate documentation of newborn resuscitation is essential for quality improvement and adherence to clinical guidelines, yet remains underutilized in practice. Previous work using 3D-CNNs and Vision Transformers (ViT) has shown promising results in detecting key activities from newborn resuscitation videos, but also highlighted the challenges in recognizing such fine-grained activities. This work investigates the potential of generative AI (GenAI) methods to improve activity recognition from such videos. Specifically, we explore the use of local vision-language models (VLMs), combined with large language models (LLMs), and compare them to a supervised TimeSFormer baseline. Using a simulated dataset comprising 13.26 hours of newborn resuscitation videos, we evaluate several zero-shot VLM-based strategies and fine-tuned VLMs with classification heads, including Low-Rank Adaptation (LoRA). Our results suggest that small (local) VLMs struggle with hallucinations, but when fine-tuned with LoRA, the results reach F1 score at 0.91, surpassing the TimeSformer results of 0.70.

---

### Momentum LMS Theory beyond Stationarity: Stability, Tracking, and Regret

Authors: Yifei Jin, Xin Zheng, Lei Guo
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.11995v2

In large-scale data processing scenarios, data often arrive in sequential streams generated by complex systems that exhibit drifting distributions and time-varying system parameters. This nonstationarity challenges theoretical analysis, as it violates classical assumptions of i.i.d. (independent and identically distributed) samples, necessitating algorithms capable of real-time updates without expensive retraining. An effective approach should process each sample in a single pass, while maintaining computational and memory complexities independent of the data stream length. Motivated by these challenges, this paper investigates the Momentum Least Mean Squares (MLMS) algorithm as an adaptive identification tool, leveraging its computational simplicity and online processing capabilities. Theoretically, we derive tracking performance and regret bounds for the MLMS in time-varying stochastic linear systems under various practical conditions. Unlike classical LMS, whose stability can be characterized by first-order random vector difference equations, MLMS introduces an additional dynamical state due to momentum, leading to second-order time-varying random vector difference equations whose stability analysis hinges on more complicated products of random matrices, which poses a substantially challenging problem to resolve. Experiments on synthetic and real-world data streams demonstrate that MLMS achieves rapid adaptation and robust tracking, in agreement with our theoretical results especially in nonstationary settings, highlighting its promise for modern streaming and online learning applications.

---

### Evaluating AGENTS.md: Are Repository-Level Context Files Helpful for Coding Agents?

Authors: Thibaud Gloaguen, Niels Mündler, Mark Müller, Veselin Raychev, Martin Vechev
Published: 2026-02-12
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2602.11988v2

A widespread practice in software development is to tailor coding agents to repositories using context files, such as AGENTS.md. Although this practice is strongly encouraged by agent developers, there is currently no rigorous investigation into whether such context files are actually effective for real-world tasks. In this work, we study this question and evaluate coding agents' task completion performance in two complementary settings: established SWE-bench tasks from popular repositories, with LLM-generated context files, and a novel collection of issues from repositories containing developer-committed context files. Surprisingly, we find that providing context files does not generally improve task success rates, while increasing inference cost by over 20% on average. This observation holds across different LLMs, coding agents, and for both LLM-generated and developer-committed context files. Specifically, we find that while instructions in the context files are well followed by coding agents, repository overviews, although popular and recommended by model providers, are not helpful. We conclude that while context files are useful for specifying non-standard coding practices, any attempts to improve performance should be rigorously evaluated before deployment.

---

### Automatic Simplification of Common Vulnerabilities and Exposures Descriptions

Authors: Varpu Vehomäki, Kimmo K. Kaski
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.11982v1

Understanding cyber security is increasingly important for individuals and organizations. However, a lot of information related to cyber security can be difficult to understand to those not familiar with the topic. In this study, we focus on investigating how large language models (LLMs) could be utilized in automatic text simplification (ATS) of Common Vulnerability and Exposure (CVE) descriptions. Automatic text simplification has been studied in several contexts, such as medical, scientific, and news texts, but it has not yet been studied to simplify texts in the rapidly changing and complex domain of cyber security. We created a baseline for cyber security ATS and a test dataset of 40 CVE descriptions, evaluated by two groups of cyber security experts in two survey rounds. We have found that while out-of-the box LLMs can make the text appear simpler, they struggle with meaning preservation. Code and data are available at https://version.aalto.fi/gitlab/vehomav1/simplification\_nmi.

---

### Spatial Chain-of-Thought: Bridging Understanding and Generation Models for Spatial Reasoning Generation

Authors: Wei Chen, Yancheng Long, Mingqiao Liu, Haojie Ding, Yankai Yang, Hongyang Wei, Yi-Fan Zhang, Bin Wen, Fan Yang, Tingting Gao, Han Li, Long Chen
Published: 2026-02-12
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.11980v1

While diffusion models have shown exceptional capabilities in aesthetic image synthesis, they often struggle with complex spatial understanding and reasoning. Existing approaches resort to Multimodal Large Language Models (MLLMs) to enhance this capability. However, they either incur high computational costs through joint training or suffer from spatial information loss when relying solely on textual prompts. To alleviate these limitations, we propose a Spatial Chain-of-Thought (SCoT) framework, a plug-and-play approach that effectively bridges the reasoning capabilities of MLLMs with the generative power of diffusion models. Specifically, we first enhance the diffusion model's layout awareness by training it on an interleaved text-coordinate instruction format. We then leverage state-of-the-art MLLMs as planners to generate comprehensive layout plans, transferring their spatial planning capabilities directly to the generation process. Extensive experiments demonstrate that our method achieves state-of-the-art performance on image generation benchmarks and significantly outperforms baselines on complex reasoning tasks, while also showing strong efficacy in image editing scenarios.

---

### Accelerating Robotic Reinforcement Learning with Agent Guidance

Authors: Haojun Chen, Zili Zou, Chengdong Ma, Yaoxiang Pu, Haotong Zhang, Yuanpei Chen, Yaodong Yang
Published: 2026-02-12
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2602.11978v2

Reinforcement Learning (RL) offers a powerful paradigm for autonomous robots to master generalist manipulation skills through trial-and-error. However, its real-world application is stifled by low sample efficiency. Recent Human-in-the-Loop (HIL) methods accelerate training by using human corrections, yet this approach faces a scalability barrier. Reliance on human supervisors imposes a 1:1 supervision ratio that limits scalability, suffers from operator fatigue over extended sessions, and introduces high variance due to inconsistent human proficiency. We present Agent-guided Policy Search (AGPS), a framework that automates the training pipeline by replacing human supervisors with a multimodal agent. Our key insight is that the agent can be viewed as a semantic world model, injecting intrinsic value priors to structure physical exploration. By using tools, the agent provides precise guidance via corrective waypoints and spatial constraints for exploration pruning. We validate our approach on three tasks, ranging from precision insertion to deformable object manipulation. Results demonstrate that AGPS outperforms HIL methods in sample efficiency. This automates the supervision pipeline, unlocking the path to labor-free and scalable robot learning. Project website: https://agps-rl.github.io/agps/.

---

### Confidence-Uncertainty Boundary Calibration for Bayesian Deep Learning in Medical Image Analysis

Authors: Hua Xu, Julián D. Arias-Londoño, Juan I. Godino-Llorente
Published: 2026-02-12
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2602.11973v2

In critical decision support systems based on medical imaging, the reliability of AI-assisted decision-making is as relevant as predictive accuracy. Although deep learning models have demonstrated significant accuracy, they frequently suffer from miscalibration, manifested as overconfidence in erroneous predictions. To facilitate clinical acceptance, it is imperative that models quantify uncertainty in a manner that correlates with prediction correctness, allowing clinicians to identify unreliable outputs for further review. To address this necessity, this paper proposes a probabilistic optimization framework grounded in Bayesian deep learning. Specifically, the Confidence-Uncertainty Boundary Curve (CUBC) is first explored as an intermediate operational target. Grounded in this target, a novel Confidence-Uncertainty Boundary Loss (CUB-Loss) is proposed to regularize the alignment between prediction confidence and uncertainty estimates during training, imposing penalties on high-certainty errors and low-certainty correct predictions. Upon completion of training optimization, a Boundary Curve Calibration Error (BCCE) metric is further introduced to measure the degree of boundary alignment in the calibrated model. Building on this measurement, a Dual Temperature Scaling (DTS) strategy is devised to perform post-hoc refinement, further adjusting the posterior predictive distribution across different confidence-uncertainty regions. The proposed framework is validated on three distinct medical imaging tasks: automatic screening of pneumonia, diabetic retinopathy detection, and identification of skin lesions. Empirical results demonstrate that the proposed approach improves uncertainty calibration across diverse modalities, maintains robust performance in data-scarce scenarios, and remains effective on severely imbalanced datasets, underscoring its potential for real clinical deployment.

---

### UPDA: Unsupervised Progressive Domain Adaptation for No-Reference Point Cloud Quality Assessment

Authors: Bingxu Xie, Fang Zhou, Jincan Wu, Yonghui Liu, Weiqing Li, Zhiyong Su
Published: 2026-02-12
Categories: eess.IV, cs.CV, cs.MM
arXiv: https://arxiv.org/abs/2602.11969v1

While no-reference point cloud quality assessment (NR-PCQA) approaches have achieved significant progress over the past decade, their performance often degrades substantially when a distribution gap exists between the training (source domain) and testing (target domain) data. However, to date, limited attention has been paid to transferring NR-PCQA models across domains. To address this challenge, we propose the first unsupervised progressive domain adaptation (UPDA) framework for NR-PCQA, which introduces a two-stage coarse-to-fine alignment paradigm to address domain shifts. At the coarse-grained stage, a discrepancy-aware coarse-grained alignment method is designed to capture relative quality relationships between cross-domain samples through a novel quality-discrepancy-aware hybrid loss, circumventing the challenges of direct absolute feature alignment. At the fine-grained stage, a perception fusion fine-grained alignment approach with symmetric feature fusion is developed to identify domain-invariant features, while a conditional discriminator selectively enhances the transfer of quality-relevant features. Extensive experiments demonstrate that the proposed UPDA effectively enhances the performance of NR-PCQA methods in cross-domain scenarios, validating its practical applicability. The code is available at https://github.com/yokeno1/UPDA-main.

---

### DHPLT: large-scale multilingual diachronic corpora and word representations for semantic change modelling

Authors: Mariia Fedorova, Andrey Kutuzov, Khonzoda Umarova
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.11968v1

In this resource paper, we present DHPLT, an open collection of diachronic corpora in 41 diverse languages. DHPLT is based on the web-crawled HPLT datasets; we use web crawl timestamps as the approximate signal of document creation time. The collection covers three time periods: 2011-2015, 2020-2021 and 2024-present (1 million documents per time period for each language). We additionally provide pre-computed word type and token embeddings and lexical substitutions for our chosen target words, while at the same time leaving it open for the other researchers to come up with their own target words using the same datasets. DHPLT aims at filling in the current lack of multilingual diachronic corpora for semantic change modelling (beyond a dozen of high-resource languages). It opens the way for a variety of new experimental setups in this field. All the resources described in this paper are available at https://data.hplt-project.org/three/diachronic/, sorted by language.

---

### Manifold-Aware Temporal Domain Generalization for Large Language Models

Authors: Yiheng Yao, Zekun Cai, Xinyuan Song, Hiroki Hill Kobayashi, Xuan Song, Ryosuke Shibasaki, Liang Zhao
Published: 2026-02-12
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.11965v1

Temporal distribution shifts are pervasive in real-world deployments of Large Language Models (LLMs), where data evolves continuously over time. While Temporal Domain Generalization (TDG) seeks to model such structured evolution, existing approaches characterize model adaptation in the full parameter space. This formulation becomes computationally infeasible for modern LLMs. This paper introduces a geometric reformulation of TDG under parameter-efficient fine-tuning. We establish that the low-dimensional temporal structure underlying model evolution can be preserved under parameter-efficient reparameterization, enabling temporal modeling without operating in the ambient parameter space. Building on this principle, we propose Manifold-aware Temporal LoRA (MaT-LoRA), which constrains temporal updates to a shared low-dimensional manifold within a low-rank adaptation subspace, and models its evolution through a structured temporal core. This reparameterization dramatically reduces temporal modeling complexity while retaining expressive power. Extensive experiments on synthetic and real-world datasets, including scientific documents, news publishers, and review ratings, demonstrate that MaT-LoRA achieves superior temporal generalization performance with practical scalability for LLMs.

---

### Gaia2: Benchmarking LLM Agents on Dynamic and Asynchronous Environments

Authors: Romain Froger, Pierre Andrews, Matteo Bettini, Amar Budhiraja, Ricardo Silveira Cabral, Virginie Do, Emilien Garreau, Jean-Baptiste Gaya, Hugo Laurençon, Maxime Lecanu, Kunal Malkan, Dheeraj Mekala et al.
Published: 2026-02-12
Categories: cs.AI
arXiv: https://arxiv.org/abs/2602.11964v1

We introduce Gaia2, a benchmark for evaluating large language model agents in realistic, asynchronous environments. Unlike prior static or synchronous evaluations, Gaia2 introduces scenarios where environments evolve independently of agent actions, requiring agents to operate under temporal constraints, adapt to noisy and dynamic events, resolve ambiguity, and collaborate with other agents. Each scenario is paired with a write-action verifier, enabling fine-grained, action-level evaluation and making Gaia2 directly usable for reinforcement learning from verifiable rewards. Our evaluation of state-of-the-art proprietary and open-source models shows that no model dominates across capabilities: GPT-5 (high) reaches the strongest overall score of 42% pass@1 but fails on time-sensitive tasks, Claude-4 Sonnet trades accuracy and speed for cost, Kimi-K2 leads among open-source models with 21% pass@1. These results highlight fundamental trade-offs between reasoning, efficiency, robustness, and expose challenges in closing the "sim2real" gap. Gaia2 is built on a consumer environment with the open-source Agents Research Environments platform and designed to be easy to extend. By releasing Gaia2 alongside the foundational ARE framework, we aim to provide the community with a flexible infrastructure for developing, benchmarking, and training the next generation of practical agent systems.

---

### Scaling Model and Data for Multilingual Machine Translation with Open Large Language Models

Authors: Yuzhe Shang, Pengzhi Gao, Wei Liu, Jian Luan, Jinsong Su
Published: 2026-02-12
Categories: cs.CL
arXiv: https://arxiv.org/abs/2602.11961v3

Open large language models (LLMs) have demonstrated improving multilingual capabilities in recent years. In this paper, we present a study of open LLMs for multilingual machine translation (MT) across a range of languages, and investigate the effects of model scaling and data scaling when adapting open LLMs to multilingual MT through continual pretraining and instruction finetuning. Based on the Gemma3 model family, we develop MiLMMT-46, which achieves top-tier multilingual translation performance across 46 languages. Extensive experiments show that MiLMMT-46 consistently outperforms recent state-of-the-art (SOTA) models, including Seed-X, HY-MT-1.5, and TranslateGemma, and achieves competitive performance with strong proprietary systems such as Google Translate and Gemini 3 Pro. Models are released at https://huggingface.co/collections/xiaomi-research/milmmt-46. Codes are released at https://github.com/xiaomi-research/gemmax.

---

### Benchmarking Vision-Language Models for French PDF-to-Markdown Conversion

Authors: Bruno Rigal, Victor Dupriez, Alexis Mignon, Ronan Le Hy, Nicolas Mery
Published: 2026-02-12
Categories: cs.CV, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2602.11960v1

This report evaluates PDF-to-Markdown conversion using recent Vision-Language Models (VLMs) on challenging French documents. Document parsing is a critical step for Retrieval-Augmented Generation (RAG) pipelines, where transcription and layout errors propagate to downstream retrieval and grounding. Existing benchmarks often emphasize English or Chinese and can over-penalize benign formatting and linearization choices (e.g., line breaks, list segmentation, alternative table renderings) that are largely irrelevant for downstream use. We introduce a French-focused benchmark of difficult pages selected via model-disagreement sampling from a corpus of 60{,}000 documents, covering handwritten forms, complex layouts, dense tables, and graphics-rich pages. Evaluation is performed with unit-test-style checks that target concrete failure modes (text presence, reading order, and local table constraints) combined with category-specific normalization designed to discount presentation-only variance. Across 15 models, we observe substantially higher robustness for the strongest proprietary models on handwriting and forms, while several open-weights systems remain competitive on standard printed layouts.

---

### RAM-Net: Expressive Linear Attention with Selectively Addressable Memory

Authors: Kaicheng Xiao, Haotian Li, Liran Dong, Guoliang Xing
Published: 2026-02-12
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2602.11958v1

While linear attention architectures offer efficient inference, compressing unbounded history into a fixed-size memory inherently limits expressivity and causes information loss. To address this limitation, we introduce Random Access Memory Network (RAM-Net), a novel architecture designed to bridge the gap between the representational capacity of full attention and the memory efficiency of linear models. The core of RAM-Net maps inputs to high-dimensional sparse vectors serving as explicit addresses, allowing the model to selectively access a massive memory state. This design enables exponential state size scaling without additional parameters, which significantly mitigates signal interference and enhances retrieval fidelity. Moreover, the inherent sparsity ensures exceptional computational efficiency, as state updates are confined to minimal entries. Extensive experiments demonstrate that RAM-Net consistently surpasses state-of-the-art baselines in fine-grained long-range retrieval tasks and achieves competitive performance in standard language modeling and zero-shot commonsense reasoning benchmarks, validating its superior capability to capture complex dependencies with significantly reduced computational overhead.

---

### Are Two LLMs Better Than One? A Student-Teacher Dual-Head LLMs Architecture for Pharmaceutical Content Optimization

Authors: Suyash Mishra, Qiang Li, Anubhav Girdhar
Published: 2026-02-12
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.11957v1

Large language models (LLMs) are increasingly used to create content in regulated domains such as pharmaceuticals, where outputs must be scientifically accurate and legally compliant. Manual quality control (QC) is slow, error prone, and can become a publication bottleneck. We introduce LRBTC, a modular LLM and vision language model (VLM) driven QC architecture covering Language, Regulatory, Brand, Technical, and Content Structure checks. LRBTC combines a Student-Teacher dual model architecture, human in the loop (HITL) workflow with waterfall rule filtering to enable scalable, verifiable content validation and optimization. On AIReg-Bench, our approach achieves 83.0% F1 and 97.5% recall, reducing missed violations by 5x compared with Gemini 2.5 Pro. On CSpelling, it improves mean accuracy by 26.7%. Error analysis further reveals that while current models are strong at detecting misspellings (92.5 recall), they fail to identify complex medical grammatical (25.0 recall) and punctuation (41.7 recall) errors, highlighting a key area for future work. This work provides a practical, plug and play solution for reliable, transparent quality control of content in high stakes, compliance critical industries. We also provide access to our Demo under MIT Licenses.

---

### TAVAE: A VAE with Adaptable Priors Explains Contextual Modulation in the Visual Cortex

Authors: Balázs Meszéna, Keith T. Murray, Julien Corbo, O. Batuhan Erkat, Márton A. Hajnal, Pierre-Olivier Polack, Gergő Orbán
Published: 2026-02-12
Categories: q-bio.NC, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2602.11956v1

The brain interprets visual information through learned regularities, a computation formalized as probabilistic inference under a prior. The visual cortex establishes priors for this inference, some delivered through established top-down connections that inform low-level cortices about statistics represented at higher levels in the cortical hierarchy. While evidence shows that adaptation leads to priors reflecting the structure of natural images, it remains unclear whether similar priors can be flexibly acquired when learning a specific task. To investigate this, we built a generative model of V1 optimized for a simple discrimination task and analyzed it together with large-scale recordings from mice performing an analogous task. In line with recent approaches, we assumed that neuronal activity in V1 corresponds to latent posteriors in the generative model, enabling investigation of task-related priors in neuronal responses. To obtain a flexible test bed, we extended the VAE formalism so that a task can be acquired efficiently by reusing previously learned representations. Task-specific priors learned by this Task-Amortized VAE were used to investigate biases in mice and model when presenting stimuli that violated trained task statistics. Mismatch between learned task statistics and incoming sensory evidence produced signatures of uncertainty in stimulus category in the TAVAE posterior, reflecting properties of bimodal response profiles in V1 recordings. The task-optimized generative model accounted for key characteristics of V1 population activity, including within-day updates to population responses. Our results confirm that flexible task-specific contextual priors can be learned on demand by the visual system and deployed as early as the entry level of visual cortex.

---

### Insights on Muon from Simple Quadratics

Authors: Antoine Gonon, Andreea-Alexandra Muşat, Nicolas Boumal
Published: 2026-02-12
Categories: math.OC, cs.LG
arXiv: https://arxiv.org/abs/2602.11948v1

Muon updates weight matrices along (approximate) polar factors of the gradients and has shown strong empirical performance in large-scale training. Existing attempts at explaining its performance largely focus on single-step comparisons (on quadratic proxies) and worst-case guarantees that treat the inexactness of the polar-factor as a nuisance ``to be argued away''. We show that already on simple strongly convex functions such as $L(W)=\frac12\|W\|_{\text{F}}^2$, these perspectives are insufficient, suggesting that understanding Muon requires going beyond local proxies and pessimistic worst-case bounds. Instead, our analysis exposes two observations that already affect behavior on simple quadratics and are not well captured by prevailing abstractions: (i) approximation error in the polar step can qualitatively alter discrete-time dynamics and improve reachability and finite-time performance -- an effect practitioners exploit to tune Muon, but that existing theory largely treats as a pure accuracy compromise; and (ii) structural properties of the objective affect finite-budget constants beyond the prevailing conditioning-based explanations. Thus, any general theory covering these cases must either incorporate these ingredients explicitly or explain why they are irrelevant in the regimes of interest.

---

### Change-Point Detection via Piecewise Linear Fitting Using MIP

Authors: Apoorva Narula, Santanu S. Dey, Yao Xie
Published: 2026-02-12
Categories: math.OC, stat.ML
arXiv: https://arxiv.org/abs/2602.11947v2

We present a new mixed-integer programming (MIP) approach for offline multiple change-point detection by casting the problem as a globally optimal piecewise linear (PWL) fitting problem. Our main contribution is a family of strengthened MIP formulations whose linear programming (LP) relaxations admit integral projections onto the segment-assignment variables, which encode the segment membership of each data point. This property yields provably tighter relaxations than existing formulations for offline multiple change-point detection. We further extend the framework to multi-dimensional PWL models with shared change-points. Extensive computational experiments on benchmark real-world datasets demonstrate that the proposed formulations achieve reductions in solution times in comparison to the state-of-the-art.

---

### Mixed-Integer Programming for Change-point Detection

Authors: Apoorva Narula, Santanu S. Dey, Yao Xie
Published: 2026-02-12
Categories: math.OC, stat.ML
arXiv: https://arxiv.org/abs/2602.11947v1

We present a new mixed-integer programming (MIP) approach for offline multiple change-point detection by casting the problem as a globally optimal piecewise linear (PWL) fitting problem. Our main contribution is a family of strengthened MIP formulations whose linear programming (LP) relaxations admit integral projections onto the segment assignment variables, which encode the segment membership of each data point. This property yields provably tighter relaxations than existing formulations for offline multiple change-point detection. We further extend the framework to two settings of active research interest: (i) multidimensional PWL models with shared change-points, and (ii) sparse change-point detection, where only a subset of dimensions undergo structural change. Extensive computational experiments on benchmark real-world datasets demonstrate that the proposed formulations achieve reductions in solution times under both $\ell_1$ and $\ell_2$ loss functions in comparison to the state-of-the-art.

---

### Towards Performance-Enhanced Model-Contrastive Federated Learning using Historical Information in Heterogeneous Scenarios

Authors: Hongliang Zhang, Jiguo Yu, Guijuan Wang, Wenshuo Ma, Tianqing He, Baobao Chai, Chunqiang Hu
Published: 2026-02-12
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.11945v1

Federated Learning (FL) enables multiple nodes to collaboratively train a model without sharing raw data. However, FL systems are usually deployed in heterogeneous scenarios, where nodes differ in both data distributions and participation frequencies, which undermines the FL performance. To tackle the above issue, this paper proposes PMFL, a performance-enhanced model-contrastive federated learning framework using historical training information. Specifically, on the node side, we design a novel model-contrastive term into the node optimization objective by incorporating historical local models to capture stable contrastive points, thereby improving the consistency of model updates in heterogeneous data distributions. On the server side, we utilize the cumulative participation count of each node to adaptively adjust its aggregation weight, thereby correcting the bias in the global objective caused by different node participation frequencies. Furthermore, the updated global model incorporates historical global models to reduce its fluctuations in performance between adjacent rounds. Extensive experiments demonstrate that PMFL achieves superior performance compared with existing FL methods in heterogeneous scenarios.

---

### Using predictive multiplicity to measure individual performance within the AI Act

Authors: Karolin Frohnapfel, Mara Seyfert, Sebastian Bordt, Ulrike von Luxburg, Kristof Meding
Published: 2026-02-12
Categories: cs.LG, cs.CY
arXiv: https://arxiv.org/abs/2602.11944v2

When building AI systems for decision support, one often encounters the phenomenon of predictive multiplicity: a single best model does not exist; instead, one can construct many models with similar overall accuracy that differ in their predictions for individual cases. Especially when decisions have a direct impact on humans, this can be highly unsatisfactory. For a person subject to high disagreement between models, one could as well have chosen a different model of similar overall accuracy that would have decided the person's case differently. We argue that this arbitrariness conflicts with the EU AI Act, which requires providers of high-risk AI systems to report performance not only at the dataset level but also for specific persons. The goal of this paper is to put predictive multiplicity in context with the EU AI Act's provisions on accuracy and to subsequently derive concrete suggestions on how to evaluate and report predictive multiplicity in practice. Specifically: (1) We introduce the AI Act's accuracy provisions and argue that incorporating information about predictive multiplicity could serve compliance with specific provisions for providers. (2) Based on this legally rigorous analysis, we suggest individual conflict ratios and $δ$-ambiguity as tools to quantify the disagreement between models on individual cases and to help detect individuals subject to conflicting predictions. (3) Based on computational insights, we derive easy-to-implement rules on how model providers could evaluate predictive multiplicity in practice. (4) Ultimately, we suggest that information about predictive multiplicity should be made available to deployers under the AI Act, enabling them to judge whether system outputs for specific individuals are reliable enough for their use case.

---