arXiv:2606.02520v1 Announce Type: new
Abstract: Subduction zones on the surface of the Earth, where abrupt sliding leads to earthquakes, are generally curved and localized. How does the geometry of these zones influence the occurrence of megathrust earthquakes? Here we use a combination of simple scaling arguments and data analysis using the differential geometry of surfaces to examine the relationship between the earthquake productivity of subduction zones and their shape. A scaling argument suggests how interface curvature changes both the accumulation and release of stress relative to planar interfaces; conformable sliding along relatively flat subduction zones should lead to rare but large events, while curved subduction zones should lead to frequent smaller events. To test this, we leverage global geometry datasets and analyze the correlation between the surface curvatures of the subduction zones and the frequency and magnitude of earthquakes therein. Our analysis shows that weakly curved slab geometries are associated with rarer larger magnitude events, while slab geometries with a larger relative dispersion in curvature are associated with frequent but smaller magnitude events. Using different scale-dependent shape metrics of the subduction zones, we show that the earthquake productivity is influenced by the conformability of the overriding and downgoing plates. More broadly, our results suggest the need to incorporate the large-scale geometry of subduction zones in computational models and predictive frameworks for earthquake risk.
Science Journals
arXiv:2605.23947v2 Announce Type: replace
Abstract: We introduce FLUID (Fountain LiqUId Delivery), a protocol that uses fountain coding and receiver feedback for low-latency delivery of data blocks over lossy networks. Idealized Automatic Repeat reQuest (ARQ) protocols are bandwidth-optimal, but must deliver every packet in a block and therefore can require additional rounds under packet loss. FLUID uses a controlled amount of slack to relax this all-packets requirement, allowing delivery to finish once enough encoded packets have been received. This yields substantially tighter delivery latency while remaining deterministically close to the ARQ bandwidth optimum.
FLUID is controlled by a slack parameter $\epsilon$. Under the Loss-Product Rule, delivery finishes once the product of packet loss fractions across transmission rounds falls below $\epsilon$. Thus, FLUID can finish delivery in a small number of rounds even when every round experiences packet loss, while $\epsilon$ controls the gap between FLUID and bandwidth-optimal ARQ.
arXiv:2602.06033v2 Announce Type: replace
Abstract: Pre-trained vision language models do not have good intuitions about the physical world. Recent work has shown that supervised fine-tuning can improve model performance on simple physical tasks. However, fine-tuned models do not appear to learn robust physical rules that can generalize to new contexts. Based on research in cognitive science, we hypothesize that models need to interact with an environment to properly learn its physical dynamics. We train models that learn through interaction with a simulated environment using reinforcement learning. While learning from interaction allows models to improve their within-task performance, it fails to produce models with generalizable physical intuitions. We find that models trained on one task do not reliably generalize to related tasks, even if the tasks share visual statistics and physical principles, and regardless of whether the models are trained through interaction.
arXiv:2606.01654v1 Announce Type: new
Abstract: Microchannel flow boiling is an efficient cooling solution for high-power-density miniaturized systems. Many studies on microchannel flow boiling focused on the dynamics of single vapor bubbles, while neglecting the interaction between bubbles, which is important in relevant applications. Here, numerical simulations are carried out to study the interaction between multiple vapor bubbles in microchannel flow boiling. The results show that for different numbers of bubbles in the microchannels with the same initial size and position of leading bubbles, the bubble size in a single-bubble microchannel is larger compared to the leading bubble of multiple-bubble cases because of heat absorption by the vaporization at the rear bubbles. As the initial volume ratio between the leading bubble and the rear bubble decreases, the leading bubble size in the downstream becomes smaller because of the reduced contact with the superheated thermal boundary layer. With increasing the Reynolds number, both the leading and the trailing bubbles increase slightly in size in the upstream of the heated region, because the bubbles at higher Reynolds number move faster and firstly get in contact with the superheated fluid. The increase in the bottom wall thickness increases the growth rate of the multiple bubble sizes with earlier bubble coalescence because of the higher upstream wall temperature by heat conduction in the solid wall.
arXiv:2605.24881v2 Announce Type: replace
Abstract: Robotic surface-interaction tasks, such as spray painting or welding, require both accurate geometric planning and precise motion execution. While modern motion planners generate valid geometric paths, they often lack the expert motor patterns observed in human operators. Conversely, learning from demonstration often tightly couples task execution to the specific training geometry, limiting transferability. We propose a modular framework that decouples geometric motion planning from execution-level expertise. Expert behavior is represented as a vocabulary of interpretable, atomic motor rules, such as velocity scaling and orientation offsets, that systematically modify a geometrically planned reference path. We train a multimodal neural network to infer rule parameters jointly from kinematic trajectory data and CAD model geometry. We evaluate our approach through dynamic simulation on L-shaped and window-shaped objects, demonstrating on simulated data that the model successfully extracts velocity and orientation rules across both topologies.
arXiv:2605.25143v2 Announce Type: replace
Abstract: Test-time scaling improves language model reasoning by spending additional compute to explore multiple solution trajectories. The key challenge is to maximize accuracy while minimizing the total number of generated tokens during reasoning. Recent PRM-guided methods score intermediate prefixes to steer this search, but most are frontier-only: they keep only the current active prefixes and irreversibly prune or resample away the rest using noisy PRM scores. This can cause premature commitment, diversity collapse, and the loss of prefixes that still admit correct continuations. We introduce stochastic backtracking over a persistent pool of historical prefixes, allowing test-time compute to revisit previously generated states instead of only expanding the current frontier. To make this efficient, we propose two complementary mechanisms. Subpool Selection strengthens greedy PRM-guided search by applying Top-N selection within random subpools, giving historical prefixes a chance to bypass over-scored frontier candidates. Power Backtrack Sequential Monte Carlo extends SMC-style resampling to the persistent pool using powered PRM scores and mixture-corrected weights. Across mathematical reasoning benchmarks and model scales, our methods consistently achieve higher accuracy per token count, and the same level of accuracy using only a fraction of the token count in comparison to strong PRM-guided baselines, demonstrating that persistent-pool stochastic backtracking provides a simple and effective way to improve the accuracy-token trade-off in test-time scaling.
arXiv:2605.07061v2 Announce Type: replace
Abstract: Joint audio-video generation models are rapidly approaching professional production quality, raising a central question: do they understand audio-visual physics, or merely generate plausible sounds and frames that violate real-world consistency? We introduce AV-Phys Bench, a benchmark for evaluating physical commonsense in joint audio-video generation. AV-Phys Bench tests models across three scene categories: Steady State, Event Transition, and Environment Transition. It covers physics-grounded subcategories drawn from real-world scenes, plus Anti-AV-Physics prompts that deliberately request physically inconsistent audio-video behavior. Each generation is evaluated along five dimensions: visual semantic adherence, audio semantic adherence, visual physical commonsense, audio physical commonsense, and cross-modal physical commonsense. Across three proprietary and four open-source models, we find that Seedance 2.0 performs best overall, but all models remain far from robust physical understanding. Performance drops sharply on event-driven and environment-driven transitions, and even strong proprietary systems collapse on Anti-AV-Physics prompts. We further introduce AV-Phys Agent, a ReAct-style evaluator that combines a multimodal language model with deterministic acoustic measurement tools, producing rankings that closely align with human ratings. Our results identify cross-modal physical consistency and transition-driven scene dynamics as key open challenges for joint audio-video generation.
arXiv:2606.02334v1 Announce Type: new
Abstract: Dataset search and reuse are strongly constrained by the quality of metadata such as natural language descriptions, which are often sparse or inconsistent. Although large language models (LLMs) can generate such descriptions automatically, little empirical guidance exists on what makes a good dataset description and what dataset context LLMs actually need. We study these questions through a literature-grounded framework of dataset description quality and a large-scale ablation study using 252 datasets (1,336 CSV files) from the European data portal data.europa.eu. We generate descriptions with LLMs in a baseline scenario and two ablation scenarios: (1) using only dataset titles, (2) titles and schema, and (3) titles, schema and representative data, and evaluate them with an LLM-as-a- judge framework and a semantic descriptive attribute analysis grounded in our quality dimensions. Our results reveal a consis- tent schema penalty: table-schemas alone often degrade narrative quality, while representative data partially restores grounding without improving overall human-facing quality. We further show that different LLMs exhibit stable descriptive personas. These findings provide practical guidance for LLM-supported data publishing workflows.
arXiv:2602.04861v2 Announce Type: replace
Abstract: Machine Learning Interatomic Potentials (MLIPs) sometimes fail to reproduce the physical smoothness of the quantum potential energy surface (PES), leading to erroneous behavior in downstream simulations that standard energy and force regression evaluations can miss. Existing evaluations, such as microcanonical molecular dynamics (MD), are computationally expensive and primarily probe near-equilibrium states. To improve evaluation metrics for MLIPs, we introduce the Bond Smoothness Characterization Test (BSCT). This efficient benchmark probes the PES via controlled bond deformations and detects non-smoothness, including discontinuities, artificial minima, and spurious forces, both near and far from equilibrium. We show that BSCT correlates strongly with MD stability while requiring a fraction of the cost of MD. To demonstrate how BSCT can guide iterative model design, we utilize an unconstrained Transformer backbone as a testbed, illustrating how refinements such as a new differentiable $k$-nearest neighbors algorithm and temperature-controlled attention reduce artifacts identified by our metric. By optimizing model design systematically based on BSCT, the resulting MLIP simultaneously achieves a low conventional E/F regression error, stable MD simulations, and robust atomistic property predictions. Our results establish BSCT as both a validation metric for practitioners to assess MLIP utility and as an "in-the-loop" model design proxy that alerts MLIP developers to physical challenges that cannot be efficiently evaluated by current MLIP benchmarks. The BSCT dataset and evaluation are available on https://github.com/ryanliu30/bsct.git
arXiv:2605.26068v3 Announce Type: replace
Abstract: Weakly supervised anomaly detection (WSAD) has developed in three primary directions: incomplete, inexact, and inaccurate supervision. However, these directions remain isolated, lacking a unified framework to assess whether they address unique challenges or share fundamental mechanisms. This paper introduces WSADBench, the first benchmark that unifies evaluation across distinct weakly supervised scenarios, benchmarking diverse approaches from specialized WSAD methods to advanced tabular foundation models. WSADBench establishes standardized protocols to evaluate 36 algorithms across 4 modalities by systematically varying label quantity, granularity, and quality, revealing the performance boundaries of various methods. Based on over 700K experiments, WSADBench reveals four critical insights: (i) Strong intrinsic correlations exist between these weak supervision scenarios, challenging the isolation of current research directions. (ii) Specialized WSAD algorithms excel only in extreme label-scarcity regimes but are quickly dominated by tabular foundation models and general classification methods as supervision increases or in OOD scenarios. (iii) Unlabeled data shows inconsistent utility across settings, with marginal gains compared to label refinement. (iv) Models exhibit asymmetric sensitivity to different types of label noise. We release WSADBench as an open-source benchmark with code and datasets to facilitate future WSAD research: https://github.com/SUFE-AILAB/WSADBench.
arXiv:2605.26436v2 Announce Type: replace
Abstract: Discrete masked diffusion language models such as LLaDA generate text through iterative denoising, where mask tokens are progressively replaced with predicted tokens. LLaDA2.1 introduced a Token-to-Token (T2T) editing mechanism that accelerates generation by directly replacing committed tokens suspected of being incorrect. However, we identify fundamental limitations of T2T editing: it couples error detection with replacement, pollutes the generation context with potentially incorrect tokens, and introduces a train-inference noise mismatch where systematic model-generated errors differ from the random perturbations seen during training. We propose Token-to-Mask (T2M) remasking, a training-free, drop-in replacement for T2T editing that resets suspected erroneous tokens back to the mask state, allowing the diffusion process to re-predict them under cleaner context. We design and empirically validate three complementary error detection strategies -- probability-based, trigger-mirrored, and temporal-difference-based -- and provide a unified theoretical analysis showing that T2M remasking purifies the generation context, converts systematic inference errors back to the model's native mask noise type, and enables delayed commitment for joint multi-position optimization. Comprehensive experiments across 12 benchmarks spanning knowledge, reasoning, mathematics, coding, and instruction following show that T2M generally improves performance on tasks requiring precise token-level output, with the largest gain on mathematics (+5.92% on CMATH). Error analysis on CMATH reveals that the dominant failure mode is last-mile token corruption -- where correct reasoning produces a corrupted final answer -- and that T2M repairs 59.4% of such cases.
arXiv:2605.28335v2 Announce Type: replace
Abstract: Federated Learning (FL) enables multiple clients to collaboratively train models without sharing raw data, but it is highly vulnerable to Byzantine attacks. Existing robust approaches can neutralize these threats but incur substantial computational overhead during high-dimensional gradient aggregation, an overhead that scales poorly with model size and increasingly dominates the training cost as modern models grow larger. To address this computational bottleneck, we propose Projected Dimensionality Reduction (PDR), a universal acceleration framework for vector-level distance-based robust aggregators, which performs robust aggregation by compressing gradients into a drastically smaller subspace via sparse random projection to efficiently compute reliability weights. This approach reduces the server computational complexity to an optimal $ \mathcal{O}(Mp) $, where $ M $ is the number of clients and $ p $ is the model dimension, matching the theoretical lower bound required merely to read the gradients. We establish convergence guarantees under standard FL assumptions in prior Byzantine-robust FL analyses. By leveraging the Subspace Embedding Theorem, we show that PDR achieves optimal convergence rates of $ \mathcal{O}(1/\sqrt{T}) $ for non-convex functions and $ \mathcal{O}(1/T) $ for strongly convex functions, where $ T $ denotes the number of iterations. Crucially, we mathematically demonstrate that this massive acceleration comes almost for free, merely inflating the inherent Byzantine error floor by a bounded, tunable factor of $ \frac{1+\epsilon}{1-\epsilon} $. Experimental results on benchmark datasets confirm that integrating PDR with existing aggregators yields orders of magnitude speedups in time efficiency while maintaining highly competitive convergence performance.
arXiv:2605.27458v2 Announce Type: replace
Abstract: Transformer has significantly propelled the development of artificial intelligence, and certainly the development of agents as well. We categorize attention structures of Transformer into two types based on the source of the input information: homogenous and heterogenous attention structures. Heterogenous attention structures, with co-attention as a typical example, process information from different sources. Heterogenous attention structure is the foundation for Transformer models to achieve more complex functions and integrate more modal information. Whether for research purposes or policy requirements, the interpretation of Transformer models with heterogenous attention structures is an important task. The fusion of information from different sources brings new challenges. Our work mainly includes two parts: method and experimentation. In terms of method, we propose an interpretation method for Transformer models with heterogenous attention structures. In terms of experimentation, based on our experimental analysis paradigm, we interpret the operating mechanisms of representative models, conduct semantic interpretation and logical interpretation.
Beyond Discreteness: Sample Complexity Analysis of Straight-Through Estimator for 1-bit Quantization
arXiv:2505.18113v2 Announce Type: replace
Abstract: Training quantized neural networks requires addressing the non-differentiable and discrete nature of the underlying optimization problem. To tackle this challenge, the straight-through estimator (STE) has become the most widely adopted heuristic, allowing backpropagation through discrete operations by introducing biased yet valid surrogate gradients. However, its theoretical properties remain largely unexplored, with few existing analyses focus on the generalization error by assuming an infinite amount of training data. In contrast, this work presents the first sample complexity analysis of STE in the context of neural network quantization. Our theoretical results highlight the critical role of sample size in the success of STE, a key insight absent from existing studies. Specifically, by analyzing the quantization-aware training of a two-layer neural network with binary weights and activations, we derive the sample complexity bounds in terms of the data dimensionality that guarantee the convergence of STE-based optimization to the global minimum for both ergodic and non-ergodic analyses. Moreover, in the presence of label noises, we prove an intriguing recurrence property of STE-gradient method, where the iterate repeatedly escape from and return to the optimal binary weights. Finally, we empirically demonstrate that STE fails for general non-Gaussian data but its effectiveness can be restored through normalization, underscoring its practical importance in effective quantization.
arXiv:2505.14411v4 Announce Type: replace
Abstract: Existing time series tokenization methods predominantly encode a constant number of samples into individual tokens. This inflexible approach can generate excessive tokens for even simple patterns like extended constant values, resulting in substantial computational overhead. Inspired by the success of byte pair encoding, we propose the first pattern-centric tokenization scheme for time series analysis. Based on a discrete vocabulary of frequent motifs, our method merges samples with underlying patterns into tokens, compressing time series adaptively. Exploiting our finite set of motifs and the continuous properties of time series, we further introduce conditional decoding as a lightweight yet powerful post-hoc optimization method, which requires no gradient computation and adds no computational overhead. On recent time series foundation models, our motif-based tokenization improves forecasting performance by 40% and boosts efficiency by 2314% on average. Conditional decoding further reduces MSE by up to 48%. In an extensive analysis, we demonstrate the adaptiveness of our tokenization to diverse temporal patterns, its generalization to unseen data, and its meaningful token representations capturing distinct time series properties, including statistical moments and trends.
arXiv:2605.27569v2 Announce Type: replace
Abstract: Machine unlearning aims to remove the influence of specific training records from a deployed model without retraining from scratch. Current protocols verify this at the output level through membership inference, retain accuracy, and forget-set accuracy, but a model can satisfy all three whilst still encoding forgotten records in its intermediate representations. We introduce RULER, a set of representation-level verification metrics. The oracle-comparative metric M2 measures whether forget-set records occupy the same representational position as in a model retrained without them. The oracle-free metric M4 detects residuals from the unlearned model's internal similarity structure alone, without retraining. Four approximate unlearning methods all pass output-level evaluation, yet under a linear mixed-effects model M2 detects significant residuals in 10 of 12 conditions (p<0.05), with effect sizes growing as the forget fraction increases. A fifth method, Bad Teacher, shows the same residuals despite a different forgetting mechanism. M4 acts as a pre-unlearning diagnostic across tabular, image, clinical text, and face-identity settings: it detects identity-level memorisation in face recognition models where no tested method fully erases the signal.
arXiv:2605.27701v2 Announce Type: replace
Abstract: We present Frost Training, a method for improving Monte Carlo-based policy optimization for a large family of LLM-as-a-judge tasks called Cross-Entropy Games. The key idea is to exploit the gradient of the reward function in embedding space. This signal is used in the Greedy Coordinate Gradient (GCG) jailbreaking technique; we demonstrate for the first time that it can also be used to boost model training. We validate our method using GRPO training for maximum-likelihood infilling. Frost Training improves the model's ability to generate high-scoring outputs, reaching higher maximum scores in a best-of-k setting, and does so at an increased speed.
arXiv:2605.14830v2 Announce Type: replace
Abstract: Agentic AI systems that autonomously perform service tasks are entering customer service operations. However, limited evidence exists on how human interventions shape service outcomes when agentic AI failures create both cognitive and emotional consequences. We study this issue through a randomized field experiment on Alibaba's Taobao platform. Workers in the treatment condition supervised an agentic AI system that resolved AI-eligible chats while continuing to handle AI-ineligible chats, whereas control workers resolved all chats without agentic AI. The findings show that AI deployment reduces average chat duration and has limited effects on retrial rates, but substantially lowers ratings for AI-eligible chats. Moreover, human intervention effectiveness in AI-eligible chats depends on the nature of AI failure, post-escalation intervention effort, and intervention timing. Human intervention preserves service quality in algorithm-triggered technical escalations, i.e., unresolved customer issues beyond the AI's capability, but is less effective in algorithm-triggered emotional escalations, i.e., where customers express frustration or dissatisfaction. These differences are partly explained by variation in workers' post-escalation intervention effort across escalation types. In algorithm-triggered emotional escalations, workers showed lower engagement: they sent fewer messages, contributed a smaller share of total chat rounds, and showed less proactivity in information seeking and solution provision. We further find that early intervention is essential for sustaining high post-escalation intervention effort. Finally, we document a positive spillover effect on AI-ineligible chats, as treated workers adapted their multitasking workflow to devote greater attention to these chats. These findings offer implications for human-in-the-loop process design in human-AI collaboration systems.
arXiv:2606.01601v1 Announce Type: new
Abstract: Visual explainability for object detection remains challenging due to the multi-instance nature of detection. Existing approaches predominantly adopt post-hoc paradigms, such as gradient-based or perturbation-based explanation methods, to interpret pretrained detectors. However, these methods require additional gradient computation or repeated model inference, resulting in limited efficiency. To address this issue, we propose an End-to-end Instance-specific Visual Explanation framework (EIVE) that directly generates instance-level saliency maps following the forward pass of Detection Transformer (DETR)-like models. Specifically, we reformulate the cross-attention mechanism in the decoder as an instance-level feature attribution pathway, so that the cross-attention of each object query corresponds to the visual attribution of its predicted instance. Based on this formulation, we design a cross-layer hybrid consensus fusion (CLHCF) module to aggregate cross-attention signals across decoder layers, producing stable and compact explanations. The explanation process of EIVE requires neither gradient computation nor input perturbation, yielding high computational efficiency, and applies to single- and multi-scale DETR-like object detectors. Finally, we present an attention-aware joint training strategy (AAJTS) as a training-oriented application, which imposes spatial constraints on cross-attention patterns to encourage stable and concentrated attribution representations, thereby improving both interpretability and detection performance. Experiments on MS COCO 2017, ExDark, and Cityscapes demonstrate that EIVE produces high-quality instance-level saliency maps and achieves performance comparable to, or better than, state-of-the-art post-hoc methods across standard metrics, while substantially improving explanation efficiency. Code is available at https://github.com/xjlDestiny/EIVE.git.
arXiv:2606.01595v1 Announce Type: new
Abstract: Bayesian inference provides a principled framework for modeling epistemic uncertainty in neural networks by treating predictions as distributions rather than deterministic values. Meanwhile, diffusion-based models for 3D molecular graph generation operate on fragile geometric structures governed by strict chemical constraints, making inference highly sensitive to uncertainty miscalibration. A largely overlooked issue is that epistemic uncertainty arising from the learned denoiser interacts with the aleatoric uncertainty intentionally injected during reverse diffusion, leading to systematic variance inflation and a mismatch between the true distribution and the simulated distribution. This effect is particularly detrimental for high-precision molecular generation, where even small deviations can violate chemical validity. In this work, we provide a theoretical and empirical analysis of how epistemic uncertainty propagates through diffusion inference and degrades sampling quality. Building on this investigation, we propose UCD (Uncertainty-Calibrated Diffusion), a simple yet effective method that calibrates the reverse diffusion process to account for epistemic uncertainty. Extensive experiments on standard 3D molecular benchmarks demonstrate that UCD consistently improves sampling quality across diverse baseline methods, establishing new state-of-the-art performance for 3D molecular diffusion. The code is available at https://github.com/jiuguaiwf/UCD.
arXiv:2606.01566v1 Announce Type: new
Abstract: Small-to-medium scientific datasets place machine learning pipelines under two compounding pressures. Single-run feature selection produces feature sets that change substantially under small perturbations of the training data, and any procedure that uses the same data for selection, tuning, and evaluation produces optimistically biased performance estimates. The two failure modes are routinely treated as separable, but in the regimes where scientific data live, they interact: an unstable selection inflates the variance of an already-optimistic score, and standard remedies for one rarely address the other. RobustModelMaker is a Python framework that couples bootstrap stability selection with strict nested cross-validation, performs all preprocessing and selection inside each fold, and produces a stability-tested feature subset together with a leakage-safe performance estimate. The framework supports nine algorithms across binary classification, multiclass classification, and regression. Behaviour is verified by a deterministic test suite spanning unit, performance, and reproducibility checks on three real scientific datasets comparing to three alternative selectors (ANOVA F-test, recursive feature elimination with cross-validation, and Boruta) on both predictive score and a Jaccard measure of selection stability. RobustModelMaker is competitive in score with the best alternative selector on each dataset, and occupies a position on the joint score-stability frontier that none of the alternatives match across all three task types. Two example applications, ovarian cancer biomarker discovery from the PLCO Trial and critical-temperature regression on the UCI Superconductivity Data, illustrate how the framework is used in practice and what trade-offs become visible when stability is treated as a first-class deliverable rather than an emergent property.
arXiv:2605.27752v2 Announce Type: replace
Abstract: LLM confidence calibration is often evaluated by comparing two signals: token-probability scores and verbalized confidence. These signals are sometimes treated as direct readouts of model uncertainty, but their comparison depends on measurement choices that are rarely made explicit. In the main analysis, we hold the verbalized-confidence elicitation fixed: a single prompt template, probability scale, and output format. We then vary the measurement axes that define the verbalized-vs-token comparison: which answer string receives the token-probability score, how that score is read from the answer tokens, and under which conditioning context it is measured. We evaluate this design on four QA benchmarks across three open 7--8B base/Instruct model families, with larger Qwen2.5 variants as same-family robustness checks. The resulting comparison is sensitive to these choices: conditioning context changes the sign or magnitude of the ECE gap across settings, token readout produces smaller but still sign-moving changes, and changing the ECE estimator has little effect. Under the default generated-answer, bare-context protocol, Instruct settings are close to parity rather than showing a large calibration gain for verbalized confidence. In a separate supplied-answer analysis, surface-plausible wrong answers receive nearly the same confidence as supplied gold answers, suggesting that verbalized confidence also reflects answer plausibility and provenance rather than correctness alone. We argue that both confidence signals should be treated as protocol-dependent behavioral measurements, and provide a reporting checklist covering elicitation provenance, scored answer, token-probability readout, and conditioning context.
arXiv:2606.01402v1 Announce Type: new
Abstract: Neural network compression is commonly achieved by pruning parameters based on local importance scores, e.g., magnitude-based pruning. We propose a complementary approach that compresses models by aggregating neurons with similar functional behavior rather than removing weights independently. Our method encodes a trained network as a polynomial ODE system and applies a lumping method called Approximate Forward Differential Equivalence to identify neurons with approximately matching induced dynamics. A single tolerance parameter, $\varepsilon$, controls the compression level and induces a smooth trade-off between model size and predictive accuracy. We evaluate the method on synthetic datasets derived from nonlinear dynamical systems with known ground-truth behavior and on public regression benchmarks. Across both settings, the proposed approach achieves substantial parameter reduction while preserving accuracy, and consistently compares favorably with magnitude-based pruning and Wanda at similar compression levels. These results suggest that differential equivalence-based aggregation is a principled and effective alternative to conventional weight-centric pruning.
arXiv:2605.01752v4 Announce Type: replace
Abstract: We study linear dueling bandits in volatile environments characterized by the simultaneous presence of post-serving contexts, delayed feedback, and adversarial corruption. Feedback is subject to unknown stochastic or adversarial delays and a cumulative corruption budget $\mathcal{C}$. To address these challenges, we propose e RCDP-UCB, which integrates a learned approximator that predicts post-serving contexts from pre-serving information. It further employs an adaptive weighting strategy that clips feature vectors to mitigate the impact of corrupted and delayed observations simultaneously. Under standard regularity conditions and a parametric post-serving mapping, we rigorously establish that our algorithm is delay-regime-agnostic, achieving a regret upper bound of $\widetilde{\mathcal{O}}(d(\sqrt{T} + \mathcal{C} + \mathcal{D}))$, where $d$ is the total feature dimension and $\mathcal{D}$ encapsulates the delay complexity. Crucially, our analysis reveals an additive cost structure between corruption and delay, avoiding the multiplicative degradation typical of prior works. We further establish lower bounds that nearly match our upper bounds up to a $\sqrt{d}$ factor for adversarial delays in the absence of post-serving contexts. Code is available at https://github.com/youngmin0oh/rcdp-public.
arXiv:2604.19532v3 Announce Type: replace
Abstract: Tokenizing music to fit the general framework of language models is a compelling challenge, especially considering the diverse symbolic structures in which music can be represented (e.g., sequences, grids, and graphs). To date, most approaches tokenize symbolic music as sequences of musical events, such as onsets, pitches, time shifts, or compound note events. This strategy is intuitive and has proven effective in Transformer-based models, but it treats the regularity of musical time implicitly: individual tokens may span different durations, resulting in non-uniform time progression. In this paper, we instead consider whether an alternative tokenization is possible, where a uniform-length musical step (e.g., a beat) serves as the basic unit. Specifically, we encode all events within a single time step at the same pitch as one token, and group tokens explicitly by time step, which resembles a sparse encoding of a piano-roll representation. We evaluate the proposed tokenization on music continuation and accompaniment generation tasks, comparing it with mainstream event-based methods. Results show improved musical quality and structural coherence, while additional analyses confirm higher efficiency and more effective capture of long-range patterns with the proposed tokenization.