Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.LG) 2026-06-18

Reliable Neural-Codec Text-to-Speech by ASR Self-Verification and Distillation: Near-Zero Catastrophic Failures Across Models and Codecs

arXiv:2606.18323v1 Announce Type: cross Abstract: Open autoregressive neural-codec text-to-speech (TTS) models sound excellent on typical inputs yet suffer stochastic catastrophic failures: on a meaningful fraction of utterances they emit silence, terminate early, or collapse into repetitive or hallucinated content. We show this failure mode is cheap to remove. Under a single format-robust metric (a catastrophic-failure rate via an ASR round-trip), best-of-N ASR self-verification drives failures to near-zero: no observed failures remain by N=2 on a standard corpus (LibriSpeech) and by N=4 on a hard prompt set. This is not an artifact of one model: the reduction replicates across four open codec-TTS systems and three neural codecs (XCodec2, SNAC, Mimi), reaching the near-zero floor by N=2 on three of the four. We then make the fix free at inference time by distilling the self-verified behaviour into the model, which recovers much of the robustness in single-shot decoding, closing ~52-58% of the failure mass on hard inputs at no test-time cost. The distillation gain concentrates where it is needed (hard inputs); on already-reliable prose there is no headroom and no detectable change. A controlled comparison adds a clean negative: offline direct preference optimization (DPO/IPO) does not beat plain supervised distillation, and an online iterative variant is promising but not statistically separable at our evaluation size. We report honestly the one model that resists (a larger Llasa where scale did not obviously help) and a rare-word capability ceiling that no self-distillation method overcomes

02.
arXiv (CS.CL) 2026-06-17

PseudoBench: Measuring How Agentic Auto-Research Fuels Pseudoscience

As Large Language Model based agents enter autonomous scientific research, their ability to resist pseudoscience becomes increasingly important. Otherwise, such systems may rapidly generate plausible yet misleading studies that contaminate academic literature and erode trust in science. We present PseudoBench, an adversarial benchmark for evaluating whether agentic auto-research systems can identify and resist pseudoscientific narratives. PseudoBench contains 200 curated pseudoscientific claim-evidence pairs across five domains and evaluates agents through an end-to-end research pipeline from experiments to writing. Testing seven state-of-the-art agents, we find that current systems readily produce persuasive reports that align with pseudoscientific premises with near-zero refusal rates and the highest resistance of only 27.4%. Stronger agents risk packaging pseudoscience in more sophisticated scientific language, increasing its apparent credibility. These findings reveal an alarming capacity to fuel pseudoscience, calling for scientific alignment before widespread deployment.

03.
arXiv (CS.AI) 2026-06-16

Multi-Granular Node Pruning for Causal Circuit Discovery

arXiv:2512.10903v2 Announce Type: replace Abstract: Circuit discovery aims to identify minimal subnetworks that are responsible for specific behaviors in large language models (LLMs). Existing approaches primarily rely on iterative edge pruning, which is computationally expensive and limited to coarse-grained units such as attention heads or MLP blocks, overlooking finer structures like individual neurons. We propose a node-level pruning framework for circuit discovery that addresses both scalability and granularity limitations. Our method introduces learnable masks across multiple levels of granularity, from entire blocks to individual neurons, within a unified optimization objective. Granularity-specific sparsity penalties guide the pruning process, allowing a comprehensive compression in a single fine-tuning run. Empirically, our approach identifies circuits that are smaller in nodes than those discovered by prior methods; moreover, we demonstrate that many neurons deemed important by coarse methods are actually irrelevant, while still maintaining task performance. Furthermore, our method has a significantly lower memory footprint, 5-10x, as it does not require keeping intermediate activations in the memory to work.

04.
arXiv (CS.CV) 2026-06-16

Beyond Scalar Rewards by Internalizing Reasoning into Score Distributions

Reward models are central to text-to-image post-training, but visual preference is subjective and better represented as a distribution over rubric scores than as a deterministic scalar. Existing scalar, score-token, and pairwise reward models over-compress uncertainty and fine-grained score differences, while reasoning-based generative rewards provide stronger judgments but are costly to deploy and difficult to use as direct optimization signals. We propose Z-Reward, a teacher-student reward modeling framework that decouples reasoning-heavy judgment from efficient reward deployment. The teacher is a large VLM that uses reasoning to infer rubric-aligned score distributions, and is trained with Group-wise Direct Score Optimization (GDSO), which combines policy-gradient rewards from distribution expectations with direct pointwise and pairwise supervision on score distributions and score gaps. The student is trained with Reasoning-Internalized Score Distillation (RISD), which transfers the teacher's reasoning-conditioned score distribution into a compact VLM without requiring explicit reasoning chains at inference time. On our internally annotated evaluation set, the 27B GDSO teacher reaches 89.6% human preference accuracy, outperforming SFT, RewardDance, and GRPO, while the 9B RISD student reaches 88.6%, outperforming the OPD baseline and closely matching the larger teacher. We further show that Z-Reward can serve as a differentiable reward signal for text-to-image optimization, yielding a 41.3% net human-preference improvement over the SFT baseline.

05.
arXiv (CS.AI) 2026-06-15

CADET: Physics-Grounded Causal Auditing and Training-Free Deconfounding of End-to-End Driving Planners

作者:

arXiv:2606.14438v1 Announce Type: cross Abstract: End-to-end (E2E) autonomous-driving planners trained by imitation are prone to statistical shortcuts: they associate scene elements that merely co-occur with expert actions (a roadside object, a building facade) with driving decisions, rather than the variables that causally determine them. Such causal confusion silently compromises reliability in long-tail scenarios, and it is difficult to detect, because prevailing open-loop metrics (L2 displacement and collision rate) are dominated by ego status and do not indicate whether a planner depends on spurious cues. Existing remedies based on causal-intervention training require retraining large models and cannot audit a planner that is already deployed. We present CADET, a training-free framework that audits, benchmarks, and repairs spurious reliance in pretrained E2E planners without any parameter update.

06.
arXiv (CS.LG) 2026-06-12

Towards One-for-All Anomaly Detection for Tabular Data

arXiv:2603.14407v2 Announce Type: replace Abstract: Tabular anomaly detection (TAD) aims to identify samples that deviate from the majority in tabular data and is critical in many real-world applications. However, existing methods follow a ``one model for one dataset (OFO)'' paradigm, which relies on dataset-specific training and thus incurs high computational cost and yields limited generalization to unseen domains. To address these limitations, we propose OFA-TAD, a generalist one-for-all (OFA) TAD framework that only requires one-time training on multiple source datasets and can generalize to unseen datasets from diverse domains on-the-fly. To realize one-for-all tabular anomaly detection, OFA-TAD extracts neighbor-distance patterns as transferable cues, and introduces multi-view neighbor-distance representations from multiple transformation-induced metric spaces to mitigate the transformation sensitivity of distance profiles. To adaptively combine multi-view distance evidence, a Mixture-of-Experts (MoE) scoring network is employed for view-specific anomaly scoring and entropy-regularized gated fusion, with a multi-strategy anomaly synthesis mechanism to support training under the one-class constraint. Extensive experiments on 34 datasets from 14 domains demonstrate that OFA-TAD achieves superior anomaly detection performance and strong cross-domain generalizability under the strict OFA setting. The source code is available at https://github.com/Shiy-Li/OFA-TAD.

07.
arXiv (CS.AI) 2026-06-16

RL-Index: Reinforcement Learning for Retrieval Index Reasoning

arXiv:2606.16316v1 Announce Type: cross Abstract: Retrieving external knowledge is essential for solving real-world tasks, yet it remains challenging when the relationship between a query and its relevant knowledge involves implicit and complex reasoning beyond surface-level semantic or lexical matching (e.g., mathematical problems relying on the same theorem or coding requiring deep reasoning). Existing approaches primarily rely on query-side reasoning (e.g., query rewriting), which introduces significant online latency and underutilizes the opportunity to perform reasoning over the knowledge corpus itself (i.e., index-side reasoning). In this paper, we propose RL-Index, an agentic indexing framework that formulates retrieval index reasoning as a reinforcement learning problem. Instead of performing reasoning at query time, RL-Index shifts reasoning to the indexing stage by augmenting documents with LLM-generated rationales that explicitly encode the latent query-knowledge relationship. To optimize the quality of these rationales, we employ Group Relative Policy Optimization (GRPO) and use retrieval similarity as a verifiable reward signal, enabling direct optimization of indexing decisions for retrieval effectiveness. Extensive experiments on the BRIGHT benchmark demonstrate that RL-Index consistently improves both retrieval and downstream question-answering performance, while significantly reducing online inference latency. Moreover, the learned rationale augmentation generalizes across diverse retrievers and generators, highlighting its robustness as a plug-and-play indexing strategy across different retrieval systems.

08.
arXiv (CS.AI) 2026-06-12

Transformer Field Theory: A Response-Theoretic Approach to Mechanistic Interpretability

arXiv:2605.25225v2 Announce Type: replace-cross Abstract: Mechanistic interpretability often studies Transformer behavior by intervening on internal activations through activation patching, causal tracing, path patching, and steering directions. This paper develops Transformer Field Theory: a response-theoretic framework in which the residual stream of a fixed forward pass is treated as a Transformer field over layer depth and token position. In this formulation, patching becomes a localized source insertion into the Transformer field, first-order sensitivity fields predict patch effects, Green functions describe downstream propagation, and patch selection is posed as an adjoint inverse problem. Empirically, we test the theory's forward response objects in GPT-2-style autoregressive Transformers. Localized Transformer-field interventions exhibit a bounded local linear regime; first-order sensitivities predict patch effects across layer-token sites; localized sources generate structured anisotropic Transformer-field propagation; high-sensitivity sites and sliced Green operators provide reduced response descriptions; and prompt-induced Transformer-field displacements partially transfer answer behavior. These results establish sensitivities, Transformer-field responses, and sliced Green operators as practical objects for organizing patching experiments, while providing the forward mathematical basis for patch-site inference and cross-scale response transfer.

09.
bioRxiv (Bioinfo) 2026-06-15

SMLMFlow: Improving Structural Resolution in Single Molecule Localization Microscopy with Flow Matching

While Single Molecule Localization Microscopy (SMLM) aims to generate precise coordinates of molecular targets in cells, the resulting point clouds are inherently blurred by additive noise sources across the experimental, imaging, and processing workflow. This blurring often limits SMLM's ability to accurately quantify complex assembled structures required to address biological issues, despite reported localization precision down to a couple of nanometers. Here, we present SMLMFlow, a machine learning framework for improving structural resolution in SMLM datasets that combines a graph neural network and a hierarchical transformer with flow matching. We show that SMLMFlow improves structural resolution and downstream quantification across different structures, including filaments and protein nano-clusters, and generalizes to new unseen photophysics models.

10.
arXiv (CS.CV) 2026-06-16

Position: The Systemic Lack of Agency in Visual Reasoning

This paper argues that a systemic lack of Agency constrains the implicit reasoning capabilities of current Vision-Language Models (VLMs). Implicit reasoning refers to the ability to autonomously discover and utilize hidden visual evidence to bridge information gaps, rather than merely relying on explicitly specified targets. This capacity underlies human visual understanding and everyday reasoning. We argue that this limitation arises from a tendency to approach visual reasoning primarily as passive semantic retrieval, rather than as active, situated reasoning that depends on autonomous visual exploration. As a result, most existing benchmarks primarily assess Passive Capacity, leaving this aspect of reasoning largely unmeasured. To address this gap, we introduce the Visual Implicit Reasoning Diagnosing Benchmark (V-IRD), which targets this missing quadrant by requiring models to derive answers strictly through autonomous visual analysis. Our results show that, despite strong retrieval abilities, prominent VLMs struggle to utilize reference objects and to attend to visual evidence that requires self-directed inquiry. Simply put, strong semantic recognition does not equate to active visual exploration, revealing a critical gap in current VLMs. More information can be found at https://haoychen.github.io/Implicit-Reasoning/

11.
arXiv (CS.AI) 2026-06-11

Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning

arXiv:2606.11634v1 Announce Type: new Abstract: The rapid progress of reasoning and agentic large language models (LLMs) has increased the demand for long-context inference, but self-attention (SA) scales quadratically with context length. To address this, we study SWARR (Sliding-Window Attention with Reinforced Adaptation for Math Reasoning), a practical recipe for adapting SWA models to mathematical reasoning. SWARR has two stages: (1) efficient conversion from a pretrained SA model to SWA with supervised fine-tuning (SFT), which avoids pretraining a new base model, and (2) policy adaptation with reinforcement learning (RL). We find that SWA still underperforms SA after SFT, and we hypothesize that this gap is caused in part by a data-architecture mismatch: most SFT data are prepared for SA models and may contain long-range dependencies that are difficult for SWA to model. Because on-policy RL optimizes self-generated trajectories under the SWA constraint, it can adapt trajectories to better match SWA. Experiments on mathematical reasoning benchmarks show that this recipe substantially narrows the gap between SWA and SA, recovering much of the accuracy lost during SWA conversion while preserving the efficiency benefits of linear-complexity attention. Our central contribution is the empirical finding that RL changes the conclusion one would draw from conversion and SFT alone about SWA's viability for math reasoning.

12.
arXiv (math.PR) 2026-06-16

Collapsibility in Multiparametric Models of Random Simplicial Complexes

作者:

arXiv:2606.15276v1 Announce Type: cross Abstract: We study collapsibility in the multiparametric models of random simplicial complexes, namely the lower and upper models. In the upper model, we improve upon a result of Farber and Nowik, and assert that the homology is a.a.s concentrated in a single dimension by proving that the complex collapses to that \di. In the lower model, we prove that the complex a.a.s collapses to the \di\ with maximal non-trivial cohomology. We then compare this threshold to the ones derived previously for the special cases of the clique complex (by Kahle) and the Linial-Meshulam model.

13.
arXiv (CS.AI) 2026-06-16

S-SPPO: Semantic-Calibrated Self-Play Preference Optimization

arXiv:2606.01561v2 Announce Type: replace Abstract: Aligning Large Language Models (LLMs) with human preferences is often formulated via Direct Preference Optimization (DPO). However, the standard Bradley-Terry instantiation of DPO is limited in modeling common departures from transitivity in human preferences. To address this, recent work has introduced Self-Play Preference Optimization (SPPO), which iteratively refines the policy by training on self-generated win-lose pairs. Our investigation, however, reveals a critical instability in SPPO: the optimization is prone to policy degeneration when the preference oracle assigns overly confident wins to semantically indistinguishable responses. To mitigate this, we propose S-SPPO, a dual-space semantic calibration framework comprising: i) Supervision Calibration via semantic gating, which anneals win rate targets toward the maximum-entropy baseline as semantic overlap increases; and ii) Representation Calibration via latent repulsion to enforce geometric diversity to prevent manifold collapse and maintain latent diversity between chosen and rejected samples. Theoretically, we show that the calibration preserves the constant-sum game structure, facilitating convergence to a Nash Equilibrium. Empirically, S-SPPO avoids the performance degradation seen in prior methods, achieving 52.19% win rate and 47.46% length-controlled win rate on AlpacaEval 2.0 with Llama-3-8B, without using additional human-annotated preferences during training. The code will be available at https://github.com/xiwenc1/s-sppo.

14.
arXiv (quant-ph) 2026-06-16

Trainable Quantum Channels as Computational Primitives for Quantum Learning

arXiv:2606.15808v1 Announce Type: new Abstract: Variational quantum learning is traditionally constrained to unitary dynamics, often treating quantum channels as detrimental noise. In this work, we reformulate the quantum channels as trainable computational primitives and establish a non-unitary quantum machine learning framework grounded in open-system dynamics. We demonstrate that the outputs of channel-enhanced quantum models form a structured superposition of multiple functional components. Each component is governed by an effective observable whose spectrum can be adaptively modulated during training, a significant departure from the spectral invariance in unitary transformations. Moreover, the proposed framework generalizes conventional unitary quantum models by retaining them as a special case while introducing additional non-unitary degrees of freedom. Furthermore, we reveal that trainable quantum channels enrich the optimization geometry through ensemble-averaged gradient and additional optimization directions induced by the Kraus operators. Empirical evaluations on classification tasks using trainable amplitude-damping and phase-damping channels confirm enhanced optimization dynamics and predictive performance. Our work provides a principled approach for leveraging quantum channels as trainable resources and advances the design of high-performance quantum learning architectures.

15.
arXiv (CS.AI) 2026-06-12

Hellinger Multimodal Variational Autoencoders

arXiv:2601.06572v4 Announce Type: replace-cross Abstract: Multimodal variational autoencoders (VAEs) are widely used for weakly supervised generative learning with multiple modalities. Predominant methods aggregate unimodal inference distributions using either a product of experts (PoE), a mixture of experts (MoE), or their combinations to approximate the joint posterior. In this work, we revisit multimodal inference through the lens of probabilistic opinion pooling, an optimization-based approach. We start from Hölder pooling with $\alpha=0.5$, which corresponds to the unique symmetric member of the $\alpha-divergence$ family, and derive a moment-matching approximation, termed Hellinger. We then leverage such an approximation to propose HELVAE, a multimodal VAE that avoids sub-sampling, yielding an efficient yet effective model that: (i) learns more expressive latent representations as additional modalities are observed; and (ii) empirically achieves better trade-offs between generative coherence and quality, outperforming state-of-the-art multimodal VAE models.

16.
arXiv (CS.AI) 2026-06-11

Noise-Guided Transport for Imitation Learning

arXiv:2509.26294v2 Announce Type: replace-cross Abstract: We consider imitation learning in the low-data regime, where only a limited number of expert demonstrations are available. In this setting, methods that rely on large-scale pretraining or high-capacity architectures can be difficult to apply, and efficiency with respect to demonstration data becomes critical. We introduce Noise-Guided Transport (NGT), a lightweight off-policy method that casts imitation as an optimal transport problem solved via adversarial training. NGT requires no pretraining or specialized architectures, incorporates uncertainty estimation by design, and is easy to implement and tune. Despite its simplicity, NGT achieves strong performance on challenging continuous control tasks, including high-dimensional Humanoid tasks, under ultra-low data regimes with as few as 20 transitions.

17.
arXiv (CS.AI) 2026-06-18

Efficient Zeroth-Order Federated Finetuning of Language Models on Resource-Constrained Devices

arXiv:2502.10239v3 Announce Type: replace-cross Abstract: Federated Learning (FL) is a promising paradigm for finetuning Large Language Models (LLMs) across distributed data sources while preserving data privacy. However, finetuning such large models is challenging on edge devices due to its high resource demand. Zeroth-order Optimization (ZO) estimates gradients through finite-difference approximations, which rely on function evaluations under random perturbations of the model parameters. Consequently, ZO with task alignment provides a potential solution, allowing finetuning using only forward passes with inference-level memory requirements and low communication overhead, but it suffers from slow convergence and higher computational demand. In this paper, we propose a new ZO-based method that applies a more efficient technique to reduce the computational demand associated with using a large number of perturbations while preserving their convergence benefits. This is achieved by splitting the model into consecutive blocks and allocating a higher number of perturbations to the second block, enabling efficient reuse of intermediate activations to update the full network with fewer forward evaluations. Our evaluation on RoBERTa-large, OPT1.3B, LLaMa-3-3.2B models shows up to $3\times$ reduction in computation compared to the other ZO-based techniques, while retaining the memory and communication benefits over first-order federated learning techniques.

18.
arXiv (CS.CV) 2026-06-11

Causal Clothes-Invariant Feature Learning for Cloth-Changing Person Re-ID

In cloth-changing person re-identification (CCReID), it is critical to learn clothes-invariant feature, which can provide discriminative ID features that remain robust against clothing changes. However, a spurious correlation currently limits existing ReID methods from effectively extracting these clothing-invariant features. This spurious correlation arises from clothing ownership: clothing is rarely shared across different identities, so models tend to memorize clothing cues for identity recognition, and this strategy generalizes poorly to unseen clothing. In this paper, we propose Causal Clothes-Invariant Learning (CCIL), which explicitly shifts CC-ReID from likelihood learning P (Y|X) to causal intervention learning P (Y|do(X)) to block the clothing shortcut. CCIL realizes this intervention through three modules: a Confounder Dictionary, an Intervention Module, and Disentangle Regularization. The causality-based modeling makes the entire model naturally clothes-invariant, effectively preventing the capture of spurious correlations in feature learning. Extensive experiments validate the effectiveness of CCIL. On PRCC and DeepChange datasets, CCIL achieves Rank-1 accuracies of 66.4% and 59.2%, outperforming state-of-the-art methods by 1.4 and 4.1 percentage points, respectively.

19.
arXiv (CS.AI) 2026-06-16

Learning Permutation Distributions via Reflected Diffusion on Ranks

arXiv:2603.17353v2 Announce Type: replace-cross Abstract: The finite symmetric group S_n provides a natural domain for permutations, yet learning probability distributions on S_n is challenging due to its factorially growing size and discrete, non-Euclidean structure. Recent permutation diffusion methods define forward noising via shuffle-based random walks (e.g., riffle shuffles) and learn reverse transitions with Plackett-Luce (PL) variants, but the resulting trajectories can be abrupt and increasingly hard to denoise as n grows. We propose Soft-Rank Diffusion, a discrete diffusion framework that replaces shuffle-based corruption with a structured soft-rank forward process: we lift permutations to a continuous latent representation of order by relaxing discrete ranks into soft ranks, yielding smoother and more tractable trajectories. For the reverse process, we introduce contextualized generalized Plackett-Luce (cGPL) denoisers that generalize prior PL-style parameterizations and improve expressivity for sequential decision structures. Experiments on sorting and combinatorial optimization benchmarks show that Soft-Rank Diffusion consistently outperforms prior diffusion baselines, with particularly strong gains in long-sequence and intrinsically sequential settings.

20.
arXiv (CS.CV) 2026-06-19

SA-VIS: Sparse frame Annotations for training Video Instance Segmentation

Recent online video instance segmentation (VIS) methods have achieved impressive results, thus becoming the preferred approach to segment instances in videos. Despite the resurgence of impressive single image models, the online (or semi-online) VIS approaches outperform single-image models (e.g., based on SAM) by using long sequences of densely annotated frames during training. However,such a training setup of VIS is expensive in the sense of compute as well as dense annotations required. In order to solve these major flaws, we argue that the effective modeling of the instances and their evolution in videos do not require densely annotated frames. To that end, we propose a simple and effective module, called Past-frames Feature Propagation (PFP) which aggregates low-dimensional features from the image encoder of multiple frames. This simple low-compute module provides tremendous learning capability in using sparse video frame labels for end-to-end training. Combined with a light-weight frame-specific Instance Queries, our Sparse frame Annotation VIS (SA-VIS) significantly improves performance over its baseline. Most interestingly, our simple design that avoids complexities effectively bridges the gap in accuracy between training on sparsely and densely annotated video sequences. This translates to a mere 0.4% drop in performance of SA-VIS when using annotations for only 1/5 of the images in the dataset. Empirically, SA-VIS shows strong improvements over the baseline on YouTube-VIS 2019/2021/2022 and Occluded VIS (OVIS) and an over 1% improvement in AP on the state-of-the-art in a limited annotations scenario.

21.
arXiv (CS.CL) 2026-06-15

LLMs Contain Multitudes: How Deployment Context Reshapes Model-Level Preferences and Values

Large language models (LLMs) are increasingly characterised in recent evaluation work as having stable, model-level preference and value systems. However, accompanying robustness checks are limited to incidental prompt perturbations such as syntax variation and option reordering. This leaves open whether the measured properties survive when the surrounding task context changes, as it does in most real deployments. We test this directly across two established pairwise paradigms: ranking country preferences and eliciting utility judgements. In both, we make the deployment context – the high-level task the model is performing while making concrete value-dependent choices – our controlled variable, varied across framings such as writing a Reddit post or a news article. Across five LLMs and over 1.2M pairwise decisions, deployment context produces variation far larger than prompt paraphrasing and temperature controls. In country preference rankings over 15 countries, context induces widespread, statistically significant rank shifts; the aggregate Global North favouritism reported in prior work is itself context-dependent, with each model's bias shifting systematically across contexts. In utility elicitation over 50 outcomes, broad cross-category ordering is preserved, but fine-grained rankings within domains vary substantially, and cardinal exchange rates between outcomes (e.g. how many lives in one region equal one in another) shift by a factor of 2.47 at the median. Reported model-level preferences and utilities are therefore better understood as context-conditioned measurements than fixed model-level properties: safety guarantees obtained under one framing provide limited assurance in another.

22.
arXiv (CS.AI) 2026-06-12

Different Layers, Different Manifolds: Module-Wise Weight-Space Geometry in Transformer Optimization

arXiv:2606.13276v1 Announce Type: cross Abstract: Weight-space geometry plays a central role in neural network optimization, yet manifold constraints are often applied uniformly across all weight matrices. In this work, we ask whether different transformer modules prefer different manifold geometries. We study Manifold Muon for GPT-2 pretraining and compare layer-wise assignments of Stiefel and DGram constraints across attention and MLP blocks. Our results show a clear asymmetry: constraining attention layers with Stiefel geometry while assigning DGram geometry to MLP layers gives the best performance among the tested configurations, whereas the inverted assignment and all-DGram configuration become unstable under the shared hyperparameter setting. We trace this failure to singular value growth in DGram-constrained attention weights, which can amplify attention logits and induce softmax saturation. These findings suggest that symmetry-aware and geometry-aware optimization for transformers should be module-specific rather than uniform.

23.
arXiv (math.PR) 2026-06-17

Full $\Gamma-$expansion for the level-two large deviation rate functionals of non-reversible one-dimensional diffusions with periodic boundary conditions

arXiv:2606.17859v1 Announce Type: new Abstract: Consider the diffusion process \begin{equation*} dX_{\epsilon}(t) = \mss b(X_{\epsilon}(t)) \, dt + \sqrt{2\, \epsilon\, \mss a(X_\epsilon(t))} \, dW_{t}, \end{equation*} on the one-dimensional torus $\bb T = [0,1)$. Here $\epsilon$ is the temperature, $W_{t}$ a Brownian motion on $\bb T$ and $\mss a$, $\mss b$ functions of class $C^{2}(\bb T)$ satisfying further conditions. Denote by $\mss P(\bb T)$ the set of probability measures on $\bb T$ equipped with the weak topology, and by $\ms I_{\epsilon}\colon \mss P(\bb T)\to [0,+\infty)$ the level two large deviation rate functional of the diffusion $X_{\epsilon}(\cdot)$. We derive a full $\Gamma-$expansion of $\ms I_{\epsilon}$, as $\epsilon \to 0$, expressing it as \begin{equation*} \ms I_{\epsilon} = \frac{1}{\epsilon} \;\ms J^{(-1)} \; +\; \ms J^{(0)} \;+\; \sum_{p=1}^{\widehat{\mf q}}\frac{1}{\theta^{(p)}_{\epsilon}}\;\ms J^{(p)}\,, \end{equation*} where $\ms J^{(-1)}$, $\ms J^{(0)}$, $\ms J^{(p)} \colon \mss P(\bb T)\to [0,+\infty]$ represent rate functionals, independent of $\epsilon$, and $\theta^{(p)}_{\epsilon}$ are the time-scales at which the Markov process $X_{\epsilon}(\cdot)$ exhibits a metastable behaviour.

24.
arXiv (CS.AI) 2026-06-15

Federated Causal Inference from Multi-Site Observational Data via Propensity Score Aggregation

arXiv:2505.17961v4 Announce Type: replace-cross Abstract: Causal inference typically assumes centralized access to individual-level data. Yet, in practice, data are often decentralized across multiple sites, making centralization infeasible due to privacy, logistical, or legal constraints. We address this problem by estimating the Average Treatment Effect (ATE) from decentralized observational data via a Federated Learning (FL) approach, allowing inference through the exchange of aggregate statistics rather than individual-level data. We propose a novel method to estimate propensity scores via a federated weighted average of local scores using Membership Weights (MW), defined as probabilities of site membership conditional on covariates. MW can be flexibly estimated with parametric or non-parametric classification models using standard FL algorithms. The resulting propensity scores are used to construct Federated Inverse Propensity Weighting (Fed-IPW) and Augmented IPW (Fed-AIPW) estimators. In contrast to meta-analysis methods, which fail when any site violates positivity, our approach exploits heterogeneity in treatment assignment across sites to improve overlap. We show that Fed-IPW and Fed-AIPW perform well under site-level heterogeneity in sample sizes, treatment mechanisms, and covariate distributions. Theoretical analysis and experiments on simulated and real-world data demonstrate clear advantages over meta-analysis and related approaches.

25.
arXiv (CS.AI) 2026-06-16

AP-GRPO: Anchor-Gated Phonetic Alignment with Policy Optimization for Pathological Speech Reconstruction

arXiv:2606.15540v1 Announce Type: cross Abstract: Pathological speech from patients with neurodegenerative and neuromotor disorders is often acoustically distorted and linguistically fragmented, making pathological speech reconstruction necessary to recover intended textual content from distorted and incomplete speech recordings. Crucially, such recordings are rarely uniformly degraded: some words or short phrases remain reliable and can serve as audible anchors for reconstructing the corrupted surrounding content. We introduce Anchor-gated Phonetic Group Relative Policy Optimization (AP-GRPO), a GRPO framework with phonetic reward that aligns speech language models (SLMs) through audible-anchor preservation and inter-anchor phonetic compatibility to the original speech signal. AP-GRPO consists of: (i) an anchor-gated reward that matches reliable audible anchors in clear regions; and (ii) an inter-anchor phonetic alignment reward that evaluates whether recovered contents are phonetically supported by the corresponding corrupted inter-anchor speech span. Across four disease conditions, AP-GRPO improves faithful speech reconstruction, and the learned anchor constraint automatically adapts to each condition and thus reveals interpretable disease-specific profiles: conditions with severe articulatory degradation require stronger anchor enforcement, whereas milder impairment or linguistically impaired conditions rely more on phonetic alignment for inter-anchor recovery.