22分钟前更新- 01TriRoute: Unified Learned Routing for Joint Adaptive Attention, Experts, and KV-Cache AllocationarXiv:2607.06601v1 Announce Type: new Abstract: Conditional computation can decouple language model quality from per-token inference cost, yet leading techniques act on a single axis in isolation: Mixture-of-Experts (MoE) sparsifies the FFN, Mixture-of-Depths (MoD) skips whole transformer blocks, and KV-cache quantization compresses attention memory. We argue these three decisions (attention resolution, expert selection, and cache bit-width) are strongly coupled and should be made jointly: a tokAndrii Balashov, Olena Ponomarova
- 02A Quiet Failure in Calibrated Virtual Screening: Marginal Conformal Prediction Under-Covers the Minority Class, and a Class-Conditional Fix Recovers ItarXiv:2607.06605v1 Announce Type: new Abstract: Conformal prediction is being adopted in drug discovery to put an honest number on model reliability: pick an error rate alpha, and the method returns prediction sets containing the true label with probability at least 1 - alpha. We show this guarantee can be dangerous on imbalanced datasets. Across four datasets, standard (marginal) conformal prediction hits its global 90% coverage target while leaving the minority class badly exposed: realized miMuhammadjon Tursunbadalov (School of Science and Technology, Champions College Prep, United States), Mustafojon Tursunbadalov (School of Science and Technology, Champions College Prep, United States)
- 03NEST: Tackling Dataset-Level Distribution Shifts via Regime-Oriented Mixture-of-ExpertsarXiv:2607.06607v1 Announce Type: new Abstract: Accurate long-term forecasting in complex systems is frequently compromised by dataset-level distribution shifts, where diverse underlying behavioral modes and evolving system states drive the dynamic multivariate time-series. While existing methods predominantly focus on local temporal shifts, they fail to explicitly model the global structural challenge where datasets are composites of distinct operational regimes. In this paper, we propose NEST,Lanhao Li, Bingshu Xie, Lijun Sun, Xin Xue, Haoyi Zhou, Jianxin Li
- 04D2PO: Optimizing Diffusion Samplers via Dynamic PreferencearXiv:2607.06609v1 Announce Type: new Abstract: We propose D2PO (Dynamic Direct Preference Optimization), a principled framework for optimizing diffusion sampling policies with respect to timestep schedules and classifier-free guidance (CFG) weights. Our work is motivated by a fundamental limitation of existing student-teacher regression frameworks; low-NFE student samplers are trained to mimic high-NFEteachers, often sacrificing high-frequency texture fidelity while preserving coarse global strJinkyu Kim, Jinyoung Choi, Bohyung Han
- 05Deep Reinforcement Learning for Reliability Based Bi-Objective Portfolio OptimizationarXiv:2607.06610v1 Announce Type: new Abstract: Portfolio optimization under uncertainty is inherently a multi-objective decision problem involving complex interactions among return, risk, market dynamics, and practical investment constraints. Existing reliability based portfolio optimization approaches primarily rely on static optimization frameworks and often fail to capture sequential decision making, tail risk, and market frictions such as transaction costs. To address these limitations, we Sounaq Das, Tanmay Sen, Raghu Nandan Sengupta, Aditya Gupta
- 06STAGformer: A Spatio-temporal Agent Graph Transformer for Micro Mobility Demand ForecastingarXiv:2607.06614v1 Announce Type: new Abstract: Accurate station-level demand forecasting is essential for the efficient operation of bike-sharing systems, yet it remains challenging due to complex spatio-temporal dependencies and the large scale of urban networks. This paper presents STAGformer, a Spatio-Temporal Agent Graph Transformer that achieves efficient global modeling with linear computational complexity. The model introduces a two-step agent attention mechanism, where a small set of leYe Zihao
- 07WHERE to Generate Matters: Budget-Aware Synthetic Augmentation for Label Skewed Federated LearningarXiv:2607.06616v1 Announce Type: new Abstract: Label skew in federated learning (FL) causes client drift and degrades global accuracy. Synthetic data augmentation can reduce this imbalance; however, full class balancing requires substantial computation cost. We propose FedEAS, a policy that assigns each client an entropy-adaptive per-class generation budget computed from its local label distribution. The budget jointly decides \emph{how much} each client generates and \emph{WHERE} the samples gSangwoo Lee, Sunghwan Park, Jaewoo Lee
- 08Inertia-1: An Open Exploration of Wearable Motion Foundation ModelsarXiv:2607.06617v1 Announce Type: new Abstract: Wearable motion sensing provides a continuous and scalable window into human behavior and health, making it a natural fit for foundation models, yet its pretraining and scaling principles remain poorly understood. Prior work studies isolated design choices, such as sensor placement or sampling frequency, often under fixed settings and narrow downstream tasks that fail to capture real-world sensing diversity. We introduce Inertia-1, a fully open expZongzhe Xu, Aakarsh Anand, Sarah Jiang, Chuntung Zhuang, Zitao Shuai, Sriram Sankararaman, Yuzhe Yang
- 09Fingerprint, Not Blueprint: How Positional Schemes Set the Default Spectral Algebra of AttentionarXiv:2607.06621v1 Announce Type: new Abstract: The pre-softmax score of an attention head is a bilinear form $score(i,j) = x_i^T M x_j$ in a learned operator $M = W_q^T W_k$. Because M is generally non-symmetric, hence non-normal, it has a complex eigenspectrum and non-orthogonal eigenvectors, the regime where non-Hermitian and random-matrix tools apply. We ask what this spectrum encodes, at three levels for previous-token and induction circuits. Statically, across seven pretrained models spannLi Hengyu (Institute for Solid State Physics, The University of Tokyo)
- 10LLM-Guided Task-Semantic Field Factorization for Industrial Process ForecastingarXiv:2607.06623v1 Announce Type: new Abstract: Process industries rely on time-series forecasting and soft sensing to estimate quality variables that are hard to measure online. Labeled data are scarce, operating regimes change frequently, and retraining models or rebuilding alignment pipelines for each scenario is costly. Such settings often provide variable tables and process documents that record variable names, units, physical meanings, and process roles. However, standard time-series backbYoucheng Zong, Runda Jia, Mingxuan Ren, Dakuo He
- 11Open-Ended Scenario Reasoning for Specialist Model AdaptationarXiv:2607.06625v1 Announce Type: new Abstract: Process industries have accumulated validated specialist models, yet sensor drift, feedstock variation, and regime switching cause these models to degrade systematically in new scenarios. Collecting new labeled data and retraining is costly, while continuing with the original model incurs persistent bias. Existing adaptation methods require modifying model parameters with sufficient labeled data, making rapid response on deployed systems difficult.Youcheng Zong, Runda Jia, Ranmeng Lin, Mingxuan Ren, Dakuo He
- 12Reward Valuation in Vision Language Models: Causal Mechanisms Underlying AnhedoniaarXiv:2607.06626v1 Announce Type: new Abstract: Recent Vision-Language Models capture increasingly complex aspects of human cognition. Here we ask whether this alignment extends to reward valuation, which we assess in a mechanistic framework built on clinical tests that were developed to evaluate anhedonia and motivational deficits in major depressive disorder. In the brain, anhedonia is frequently linked to dysregulation in the Nucleus Accumbens (NAc) and the broader dopaminergic reward system.Melika Honarmand, Samin Mahdipour Aghabagher, Martin Schrimpf
- 13Cross-Trajectory Chimera Interventions Reveal Dissociable Roles of Weight Magnitude and Direction in GrokkingarXiv:2607.06628v1 Announce Type: new Abstract: Which properties of a partially trained network are causally portable to a different, independently trained network? Single-trajectory interventions show necessity within one run, not portability across runs. We introduce cross-trajectory chimera interventions: given two runs from different seeds, we split each weight vector into a norm and a unit direction, recombine one run's norm with the other's direction, and continue training. On two modular-Truong Xuan Khanh
- 14STST-JEPA: Shallow-Target Spatio-Temporal Joint Embedding Prediction Architecture For EEG Self-Supervised LearningarXiv:2607.06629v1 Announce Type: new Abstract: Brain age -- the age inferred from a physiological recording -- is an emerging biomarker whose deviation from chronological age tracks neurological and psychiatric burden, and EEG is an attractive substrate for it because it is cheap, portable, and temporally rich. Yet EEG brain-age models must contend with cross-site montage heterogeneity, small labelled cohorts, and dominant subject-level non-stationarity, and few EEG foundation models have been Roy Segal, Yoni Svechinsky, Tomer Fekete
- 15When Certificates Fail: A Unified Safety Framework for Embedded Neural Interface ModelsarXiv:2607.06630v1 Announce Type: new Abstract: Formal robustness certificates for embedded neural-interface models can pass while task accuracy collapses: at perturbation budget e=0.25, EEGNet classification accuracy drops by 25.7% under projected-gradient attack while the Lipschitz-style certificate remains valid for all 9 tested subjects. We argue that this gap between mathematical certification and operational safety is one instance of a broader alignment failure in neural interfaces, where Jasmeet Singh Bindra
- 16Does Demand Response Increase Vulnerability to Cyber Attacks by Adversarial Data Modifications?arXiv:2607.06632v1 Announce Type: new Abstract: Adversarial attacks are crafted data manipulations that aim to deteriorate the outcomes of prediction or decision-making algorithms. In the energy systems literature, adversarial attacks have been studied with a focus on problems regarding the electricity grid. Such problems include forecasting and grid state estimation, where adversarial attacks are also known as false data injection attacks. Only few studies have analyzed the potential impact thaClemens Kortmann, Eike Cramer
- 17When Do Geometric Algebra Layers Beat Scalarization? A Controlled Study on SO(3)-Equivariant Vector LawsarXiv:2607.06634v1 Announce Type: new Abstract: Compact networks built from Clifford algebra Cl(3,0) primitives are exactly SO(3)-equivariant and learn synthetic 3D vector laws from few samples. We ask whether the geometric algebra structure itself contributes anything beyond exact equivariance. We compare against a minimal scalarization baseline: invariant dot products fed to a small MLP that outputs coefficients on the equivariant basis {v_i, v_i x v_j}, which is also exactly equivariant. On sFabien Polly
- 18Optimized Instance Alteration for Explaining and Assessing Robustness of ClassifiersarXiv:2607.06637v1 Announce Type: new Abstract: In this work, we propose a unified approach for diagnosing misclassification and assessing the robustness of black-box classifiers. Central to our method is an optimization framework that modifies an instance so that the classifier predicts a specified target label, while ensuring that the modification remains easily explainable. The objective function contains two components: an explainability-aware $L_0$ (XA-$L_0$) penalty that promotes sparse anEvgenii Kuriabov, David Miller, Jia Li
- 19UASPL: Uncertainty-Aware Self-Paced Learning with Evidential Neural NetworksarXiv:2607.06638v1 Announce Type: new Abstract: Self-paced learning (SPL) is an effective learning paradigm that simulates the human learning process by progressing from easy to difficult samples based on the value of the loss function during the learning process. It has shown great potential in improving model performance and training efficiency. However, the prediction results of samples with smaller loss values are not necessarily reliable, indicating that such samples are not always simple sYifan Zhang, Yuxin Hu, Zhuobin Hao, Xiaozhuan Gao, Lipeng Pan
- 20At-Grok Is Not Converged:A Measurement-Validity Audit for Grokking Representation MetricsarXiv:2607.06639v1 Announce Type: new Abstract: On modular arithmetic, a network's embedding keeps compressing for tens of thousands of steps after it has already generalized. Reading effective rank at the grokking transition overstates the converged value by 3-5x on an MLP, and by 1.3-1.5x on a transformer trained to convergence; on the MLP it also erases which cells compress at all. Compression lags the accuracy transition by an amount on the order of the time-to-grok, at least 10,000 steps, rTruong Xuan Khanh
- 21The Rank-One Corner: How Much Value Equivalence Does a Task Need from a World Model?arXiv:2607.06640v1 Announce Type: new Abstract: A learned world model is usually judged by how faithfully it reconstructs its observations or predicts reward, as though quality were something the model simply has or lacks. But what a task actually needs from a model is narrower: the few predictive coordinates its queries depend on, which we call the closure. We show that how much of that closure a latent comes to represent is set not by the model's capacity or its observations but by the dimensiDonna Vakalis
- 22The Approximation Ratio for the Risk of Myopic Bayesian Active Learning for Linear RegressionarXiv:2607.06642v1 Announce Type: new Abstract: Active learning studies the fundamental question: what data should we choose to observe? The greedy algorithm in optimal experiment design is a common heuristic and also equivalent to myopic Bayesian active learning for linear regression, the common framework where long-term planning is replaced with the one-step optimal choice. In this work, we prove a first-of-its-kind approximation ratio for the greedy algorithm's risk that is tight up to an absStephen Mussmann
- 23Diffusion enabled Optimal Transport distances for graph matchingarXiv:2607.06646v1 Announce Type: new Abstract: This paper introduces Diffusion Semi-Relaxed Fused Gromov-Wasserstein (DsrFGW), a novel method for graph comparison that unifies node features and structural connectivity through optimal transport. While traditional Gromov-Wasserstein and semi-relaxed variants (srGW, srFGW) capture graph structure, they often struggle with sparse, noisy, or partially observed graphs. Inspired by Graph Diffusion Distance, which posits graphs are similar if they enabIman Seyedi, Francesco Archetti
- 24Final Checkpoints Are Not Enough: Analyzing Latent Reasoning Faithfulness Along Training TrajectoriesarXiv:2607.06648v1 Announce Type: new Abstract: Latent reasoning methods perform multi-step inference entirely in the model's continuous hidden states, promising more compact and efficient reasoning. However, these opaque hidden states raise a question of faithfulness: whether these latent reasoning steps causally drive the final answer. Prior work investigates this question at converged checkpoints and reports several unfaithful behaviors, such as latent reasoning steps that can be replaced witHengyu Jin, Shu Yang, Di Wang
- 25Entropy-Guided Tensor Compression for Multimodal Federated Learning on Edge DevicesarXiv:2607.06651v1 Announce Type: new Abstract: Federated learning (FL) over mobile and edge devices increasingly involves multimodal models in which clients differ in both sensing capability and computational capacity. Existing update compression schemes typically apply uniform policies across layers and devices, without accounting for modality-specific differences in spectral structure and compressibility. We propose MESH-FL, an entropy-guided matrix product state (MPS) update-compression framQuoc Bao Phan, Tuy Tan Nguyen
- 26From Jumps to Signatures: a Generative Method for Temporal Point ProcessesarXiv:2607.06652v1 Announce Type: new Abstract: Rough path signatures are a universal feature map for continuous paths and, via the expected signature, characterise path distributions. These guarantees do not directly extend to cadlag paths of Temporal Point Processes (TPPs), limiting the use of signature methods for event sequences. Furthermore, neural TPP models, including recent generative approaches, optimise per-event objectives with no global sequence-level loss, while evaluation of variabNiels Cariou-Kotlarek, Vasileios Lampos
- 27Dual Attention Heads for Personalized Federated Learning in ECG ClassificationarXiv:2607.06653v1 Announce Type: new Abstract: Federated learning (FL) enables collaborative model training across institutions without sharing sensitive patient data. However, the inherent heterogeneity of electrocardiogram (ECG) data across healthcare providers presents significant technical challenges for robust classification. We propose FedDualAtt, a personalized federated learning approach that splits transformer attention heads into global and local branches. Global heads are aggregated Kien Le, Joseph Lindley, Quoc Bao Phan, Tuy Tan Nguyen
- 28Robust Human-AI Complementarity under UncertaintyarXiv:2607.06656v1 Announce Type: new Abstract: Machine learning models are often intended to augment rather than replace human decision makers, by providing information that is complementary to human judgement. Yet, in practice, human decision makers routinely fail to realize such complementary gains, even when models provide useful signal. In this work, we study how asymmetric information about the quality of information available to a human decision maker vs. an AI impacts the ability of a deYewon Byun, Bryan Wilder
- 29Trees from Marginals: Autoregressive drafting with factorized priorsarXiv:2607.06763v1 Announce Type: new Abstract: Speculative decoding greatly increases the interactivity of autoregressive language models by trading off computation for extra tokens generated in a single forward pass. Factorized draft models are especially efficient because they predict future-token marginals in parallel, but their independence assumption causes acceptance rates to degrade sharply as the speculative budget grows. We analyze this limitation and introduce Weaver, a lightweight auYuma Oda, Ryan Mathieu, Roman Knyazhitskiy, Artur Chakhvadze
- 30Efficient Long-Horizon Learning for Learned OptimizationarXiv:2607.06772v1 Announce Type: new Abstract: Learned optimization aims to improve upon hand-designed optimizers (e.g., Adam and Muon) by meta-learning small neural network optimizers over a distribution of tasks. While recent work has greatly advanced the architectural design and inductive biases of learned optimizers (LOs), current meta-training approaches still suffer from two main difficulties: (1) they cannot efficiently scale meta-training to long-horizon inner problems and (2) they ofteXiaolong Huang, Benjamin Th\'erien, James Harrison, Eugene Belilovsky