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.CL) 2026-06-17

findsylls: A Language-Agnostic Toolkit for Syllable-Level Speech Tokenization and Embedding

Syllable-level units offer compact and linguistically meaningful representations for spoken language modeling and unsupervised word discovery, but research on syllabification remains fragmented across disparate implementations, datasets, and evaluation protocols. We introduce findsylls, a modular, language-agnostic toolkit that unifies classical syllable detectors and end-to-end syllabifiers under a common interface for syllable segmentation, embedding extraction, and multi-granular evaluation. The toolkit implements and standardizes widely used methods (e.g., Sylber, VG-HuBERT) and allows their components to be recombined, enabling controlled comparisons of representations, algorithms, and token rates. We demonstrate findsylls on English and Spanish corpora and on new hand-annotated data from Kono, an underdocumented Central Mande language, illustrating how a single framework can support reproducible syllable-level experiments across both high-resource and under-resourced settings.

02.
arXiv (CS.AI) 2026-06-16

Green SARC: Predictive Cost and Carbon Governance for Agentic AI Systems

arXiv:2606.15954v1 Announce Type: cross Abstract: Agentic AI systems act through tools and sub-agents, yet the controls meant to bound their financial and environmental cost still sit on dashboards evaluated beside or after execution. Green SARC applies the SARC governance-by-architecture framework – four enforcement sites in the agent loop – to FinOps and GreenOps, contributing the theory of what to enforce and how to predict it. We report four policy-independent results. (i) The unconstrained "State Snowball" is $\Theta(n^2)$ in loop depth; on 3,000 real multi-step plans (SWE-rebench) it holds on 100%, with median curvature $\hat{c}_2=216$ exceeding the linear-accretion prediction $p/2=134$ – real plans accrete faster than the model. (ii) On real residuals the Normal-$\sigma$ gate under-covers (92% at nominal 95%); split-conformal calibration holds (95.2%). (iii) A soft Lagrangian penalty tuned to the budget in expectation breaches it on 91.5% of seeds; the architectural gate breaches 0%. (iv) Under binding budgets the gate's over-budget incidence is 0% on synthetic and real (BurstGPT) arrivals. End-to-end token/USD/carbon savings (47–55%) are real but policy-dependent in magnitude – set by a scope-cap knob, not by gate rejections. The library is open-source, dependency-free, and ships a regeneration script for every cited number.

03.
arXiv (CS.CV) 2026-06-12

Skeleton Sparsification and Densification Scale-Spaces

The Hamilton-Jacobi skeleton, also known as the medial axis, is a powerful shape descriptor that represents binary objects in terms of the centres of maximal inscribed discs. Despite its broad applicability, the medial axis suffers from sensitivity to noise: Minor boundary variations can lead to disproportionately large and undesirable expansions of the skeleton. Classical pruning methods mitigate this shortcoming by systematically removing extraneous skeletal branches. This sequential simplification of skeletons resembles the principle of sparsification scale-spaces that embed images into a family of reconstructions from increasingly sparse pixel representations. We combine both worlds by introducing skeletonisation scale-spaces: They leverage sparsification of the medial axis to achieve hierarchical simplification of shapes. Unlike conventional pruning, our framework inherently satisfies key scale-space properties such as hierarchical architecture, controllable simplification, and equivariance to geometric transformations. We provide a rigorous theoretical foundation in both continuous and discrete formulations and extend the concept further with densification. By growing the skeleton successively instead of shrinking it, we allow inverse progression from coarse to fine scales. Densification scale-spaces can even reach beyond the original skeleton to produce overcomplete shape representations with relevancy for practical applications. Through proof-of-concept experiments, we demonstrate the effectiveness of our framework for practical tasks including robust skeletonisation, shape compression, and stiffness enhancement for additive manufacturing.

04.
arXiv (CS.CV) 2026-06-12

Acquisition state behaves as a structured, measurable variable governing lung-nodule AI: kernel-driven measurement instability and noise-driven detection fragility, invisible to DICOM metadata

AI governance for medical imaging is formalizing: the 2026 ACR-SIIM Practice Parameter recommends local acceptance testing and ongoing drift monitoring, and the ACR Assess-AI registry monitors AI outputs using DICOM metadata for context. We argue that a necessary, currently unmonitored layer sits beneath output metrics: whether incoming studies remain within the acquisition envelope a model was validated on. Using a LUNA16-trained MONAI RetinaNet lung-nodule detector, we test whether acquisition state behaves as a structured, measurable variable. On real paired CT differing only in reconstruction kernel (NLST B30f vs B80f), kernel alone shifted AI-measured diameter and flipped a Fleischner size category in 5.2% (8 of 155) of nodules at fixed patient and acquisition, while detection confidence was unchanged (Wilcoxon p=0.22). Under controlled LIDC-IDRI perturbations the effects dissociated by axis: the noise axis degraded detection confidence (p=5.9e-32, concentrated in nodules under 6 mm) but not measurement, while the frequency/kernel axis corrupted measurement (p=8.6e-13) but not detection. A 4-feature pixel fingerprint recovered reconstruction identity (patient-level AUC about 0.95 on real CT, 0.995 on a QIBA phantom) where the ConvolutionKernel DICOM tag was uninformative (identical labels across reconstructions). The kernel axis transported across four manufacturers (leave-one-vendor-out AUC 0.94-0.98, matching the within-vendor ceiling). Acquisition state thus maps to distinct AI failure modes, frequency content to measurement reliability and noise to detection sensitivity, and is not recoverable from metadata. Acquisition-aware, input-side validation is the missing layer for the acceptance-testing and drift-monitoring requirements now entering imaging-AI accreditation.

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

RAMAC: Multimodal Risk-Aware Offline Reinforcement Learning and the Role of Behavior Regularization

arXiv:2510.02695v3 Announce Type: replace-cross Abstract: In safety-critical domains where online data collection is infeasible, offline reinforcement learning (RL) is attractive only if policies achieve high returns without catastrophic lower-tail risk. Prior work on risk-averse offline RL achieves safety at the cost of either (i) value/model-based pessimism or (ii) restricted policy classes that limit expressiveness, whereas diffusion/flow-based expressive generative policies have largely been used in risk-neutral settings. We introduce Risk-Aware Multimodal Actor-Critic (RAMAC), a simple, modular, model-free framework that couples an expressive generative actor (e.g., diffusion/flow) with a distributional critic and optimizes a composite objective that combines Conditional Value-at-Risk (CVaR) with behavioral cloning (BC), enabling risk-sensitive learning in complex multimodal scenarios. Since out-of-distribution (OOD) actions are a major driver of catastrophic failures in offline RL, we further provide an objective-level analysis showing that controlling behavior divergence via BC suppresses OOD actions and stabilizes CVaR. Instantiating RAMAC with a diffusion actor, we illustrate these insights on a 2-D risky bandit and evaluate on Stochastic-D4RL, observing consistent gains in $\mathrm{CVaR}_{0.1}$ while maintaining strong returns. The code and experimental results are available on the \href{https://kaifukazawa.github.io/ramac-project/} {project website}

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

SkillAudit: Ground-Truth-Free Skill Evolution via Paired Trajectory Auditing

arXiv:2606.14239v1 Announce Type: new Abstract: Agent skills are structured procedural packages that guide frozen LLM agents in specialized workflows. Skills rarely remain sufficient after deployment: edge cases, API changes, and deployment constraints become visible only through use, making skill evolution a practical necessity. Existing methods depend on privileged feedback such as held-out validation scores, hidden test outcomes, or environment rewards – signals often unavailable when a practitioner has only a task description and workspace data. We introduce SkillAudit, a framework for evolving agent skills without ground-truth feedback. The key idea is paired trajectory auditing: at each iteration, the same task is executed with and without the candidate skill, isolating how the skill changes agent behavior without external labels. To turn behavioral differences into edit guidance, SkillAudit uses Process-Aligned Contrastive Evaluation (PACE), a cluster of evaluators that maps trajectory divergences to diagnostic signals linked to specific passages in the skill document. A structural verifier, compiled once from the task specification and then fixed, checks task constraints and rolls back harmful updates. SkillAudit routes edits through two pipelines: Refine removes noisy or irrelevant guidance from broadly useful skills, while Repair replaces passages that conflict with the task. Across 89 containerized tasks spanning 8 professional domains, SkillAudit achieves 73.9% average task reward, outperforming an agent without skills (40.9%) and the static expert skill (56.7%). These gains are obtained without accessing hidden tests, reference solutions, or external scoring functions during evolution.

07.
arXiv (CS.AI) 2026-06-15

STREAM: Multi-Tier LLM Inference Middleware with Dual-Channel HPC Token Streaming

arXiv:2606.13968v1 Announce Type: cross Abstract: Researchers and practitioners working with large language models face a fragmented landscape: local models are free and private but hardware limits the model size and context windows a researcher can use; institutional HPC centers offer powerful GPU resources at no marginal cost and keep data within institutional boundaries, but operate behind firewalls and are designed for batch jobs rather than interactive use; commercial cloud APIs provide frontier-model quality on demand but impose significant cost and data retention policies unsuitable for sensitive research data. No existing system unifies all three. STREAM (Smart Tiered Routing Engine for AI Models) addresses this gap with four contributions: (1) a three-tier routing architecture combining local, HPC, and cloud inference with a local LLM-based complexity judge; (2) a dual-channel HPC streaming architecture that separates the Globus Compute control plane (authentication and job dispatch) from a WebSocket relay data plane (token delivery), enabling sub-second TTFT (0.54 s median, 21.1x over batch mode's 11.40 s) through institutional firewalls without VPN or firewall rule changes, with end-to-end AES-256-GCM encryption ensuring the relay operator cannot read token payloads; (3) tier-aware context summarization that prevents long conversations from forcing simple queries onto expensive tiers; and (4) an HPC-as-API proxy mode that exposes HPC inference as an OpenAI-compatible endpoint callable from any standard client with no HPC expertise, a deployment pattern made practical only by the sub-second TTFT of contribution (2). Llama 3.2 3B achieves 85.1% free-tier retention on a 1,200-query benchmark spanning ten domains. Measured TTFT: 0.26 s local, 0.54 s HPC (relay), 1.68 s cloud.

08.
arXiv (CS.LG) 2026-06-18

Risk Stratification for ICU Delirium using Pervasive Ambient Sensing Information

arXiv:2606.19292v1 Announce Type: new Abstract: Delirium is a common and serious complication in the Intensive Care Unit (ICU), associated with increased morbidity, prolonged hospital stays, and higher healthcare costs. Despite its prevalence, early prediction and prevention remain challenging. Environmental factors such as ambient sound and light may influence the onset of delirium, yet they are often overlooked in risk assessments. In this study, we examined whether light intensity and sound pressure levels can independently predict delirium across multiple prediction horizons. We evaluated four efficient sequential neural network models on data collected from 9 ICUs across 309 patients to predict delirium for 10 prediction-window sizes. We reported feature importance and direction of influence using Shapley Additive Explanations analysis. The convolutional model achieved the strongest discrimination, with AUC = 0.80 on sound data and on combined data. Sound features were the dominant predictors overall. Integrating sound with light improved short-term ($

09.
arXiv (quant-ph) 2026-06-19

Quantifying Entanglement via Quantum Wasserstein Distances

arXiv:2606.04969v2 Announce Type: replace Abstract: We propose a bipartite entanglement measure defined as the minimal order-1 quantum Wasserstein distance from a state to the set of separable states. Owing to the universal data-processing inequality of the Wasserstein metric, the measure satisfies all fundamental axioms within a single geometric framework. A Lipschitz dual formulation yields explicit lower bounds for pure and mixed states, a sharp constant for two-qubit systems, and an expected value for Haar-random pure states. We further establish a quantitative connection to entanglement witnesses: any negative witness expectation value certifies a lower bound, and the dual variational bound is exactly the maximal violation achievable by a Lipschitz-1 witness. The approach naturally provides subadditivity, trace-distance estimates, and bounds on local observables, while pointing toward large-deviation conjectures. This work introduces a framework at the interface of entanglement theory, optimal transport, and experimental entanglement detection.

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

An Exploratory Study of Blood Glucose Estimation from Photoplethysmography Signals using Machine Learning

arXiv:2606.15927v1 Announce Type: new Abstract: Diabetes and extreme blood sugar levels are some of the major health problems faced by humans today across the world. While Continuous Glucose Monitoring (CGM) has emerged as an effective technology for management of diabetes as well as for monitoring blood sugar levels, this technology has traditionally been invasive (that is, requiring the piercing of the skin) and carries the risk of irritation, induration, etc. This highlights the need for accurate and non-invasive CGM methods that can be deployed at scale. With the emergence of various sensing technologies and their integration in wearables like the smart-watch, we now have the capability to continuously monitor body signals like the Photoplethysmogram (PPG) in a non-invasive manner. Having the ability to continuously monitor blood glucose through CGMs and continuously monitor PPG signals through a smart-watch offers an opportunity to get dense data on these two, opening the possibility of building machine learning and deep learning based models to estimate blood glucose level from PPG signals. In this work, we first present a paired dataset comprising continuous PPG signals from a smartwatch along with glucose values recorded using a CGM device. We also present the results of some preliminary experimental explorations performed on our dataset. These preliminary results suggest that some predictive signals may exist, though more exploration is needed with more data from a larger number of individuals. The dataset can be accessed at https://zenodo.org/records/20577959

12.
arXiv (CS.LG) 2026-06-16

Beyond Artifacts: Towards Generalizable Synthetic Song Detection via Music-Intrinsic Features

arXiv:2606.16612v1 Announce Type: cross Abstract: The rapid advancement of AI music generators highlights the urgent need for reliable Synthetic Song Detection (SSD). Existing SSD methods often rely on low-level artifacts or fixed feature assumptions, struggling to capture generator-agnostic cues. To address this, we propose Sofia (Synthetic-song detection framework via music features), a flexible framework that models music-intrinsic attributes via feature-specific experts and an adaptive Mixture-of-Experts (MoE) module. By configuring Sofia with representative Vocal, Audio-effect, Global structure features, and their combinations, we present their individual and complementary contributions. To comprehensively evaluate our framework, we further construct MUSIC8K, a challenging benchmark featuring lastest emerging generators and realistic audio perturbations. Experiments show that Sofia learns generator-agnostic representations from music-intrinsic features, improving the F1 score by 18.5 points over the strongest baseline on MUSIC8K-O while maintaining strong robustness.

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

RetailBench: Benchmarking long horizon reasoning and coherent decision making of LLM agents in realistic retail environments

arXiv:2606.15862v1 Announce Type: new Abstract: Large language model (LLM) agents have made rapid progress on short-horizon, well-scoped tasks, yet their ability to sustain coherent decisions in dynamic long-horizon environments remains uncertain. We introduce RetailBench, a data-grounded simulation benchmark for evaluating tool-using LLM agents in single-store supermarket operation. RetailBench models retail management as a partially observable decision process and is designed to support thousand-day-scale simulations. In this environment, agents must manage pricing, replenishment, supplier selection, shelf assortment, inventory aging, customer feedback, external events, and cash-flow constraints. We evaluate seven contemporary LLMs under representative agent frameworks over a 180-day evaluation horizon and compare them with a privileged oracle policy. Results show substantial variation across models: only a small subset survives the full evaluation horizon, and even the strongest LLM runs remain substantially behind the oracle policy in final net worth and sales outcomes. Behavioral analysis attributes these gaps to incomplete evidence acquisition, surface-level decision making, and the lack of a consistent long-horizon policy. RetailBench provides a controlled testbed for studying reliable autonomy in economically grounded long-horizon decision-making.

14.
arXiv (CS.AI) 2026-06-11

On the Geometry of On-Policy Distillation

arXiv:2606.07082v2 Announce Type: replace-cross Abstract: On-policy distillation (OPD) is increasingly used to improve large language model reasoning, but its training dynamics remain poorly understood. We characterize the trajectory of OPD updates in parameter space and compare it with supervised fine-tuning (SFT) and reinforcement learning with verifiable rewards (RLVR). A suite of parameter-space diagnostics consistently places OPD in a relaxed off-principal regime: compared with SFT, its updates affect fewer weights and avoid principal directions more strongly, while compared with RLVR, they remain less tightly constrained. Beyond this static localization, OPD exhibits subspace locking: its cumulative updates rapidly enter a narrow low-dimensional channel. Constraining training to the update subspace formed early in training preserves OPD performance but substantially degrades SFT, indicating that the locked subspace is functionally sufficient for OPD. Control experiments further show that sparsifying the update tokens and shifting rollout generation off-policy preserve the rank dynamics, whereas mixing the OPD objective with RLVR changes them. Overall, these results suggest that OPD is not merely an intermediate point between SFT and RLVR, but induces its own update geometry in parameter space.

15.
arXiv (CS.AI) 2026-06-16

Reward Hacking in Language Model Agents: Revisiting AI Safety Gridworlds

arXiv:2606.15385v1 Announce Type: new Abstract: Reward hacking, where AI systems exploit misspecified objectives to achieve high reward without satisfying intended goals, remains a central challenge in AI safety. Yet most known instances have been discovered post hoc in frontier systems where controlled study is impractical. We adapt the AI Safety Gridworlds framework into a text-based evaluation suite that reformulates classic reinforcement learning safety tasks for language-based agents. Across frontier and mid-scale models, we find that specification gaming emerges zero-shot: models systematically achieve high observed reward while underperforming on hidden safety objectives, and even apparently safe behaviors can reflect misunderstanding rather than principled safety. Reinforcement learning does not correct these failures: direct reward optimization widens the gap between observed and hidden reward, as the model's initial competence causes it to lock into locally rewarding strategies before discovering safer alternatives. This pattern persists across model scales (1.5B–14B) and is not resolved by finer credit assignment, exploration prompts, or entropy regularization. Our results show that reward hacking arises naturally when optimizing proxy objectives with capable language model agents and resists standard mitigations, suggesting that proxy-reward failures in agentic settings may require approaches beyond standard exploration and credit-assignment fixes. To facilitate reproducibility, the code for this work is available at \href{https://github.com/asparius/verl-agent-safety}{our public repository}.

16.
arXiv (CS.CL) 2026-06-18

PEC-Home: Interpretation of Progressively Elliptical Commands in Smart Homes

Recent advancements in Large Language Models (LLMs) have empowered home assistants with natural language interaction capabilities. However, current assistants overlook the progressive omission that occurs in human dialogue as shared context accumulates, leading to more elliptical expressions for efficient communication. Thus, current assistants still struggle to interpret such elliptical expressions accurately, which limits their effectiveness in real-world applications. In practical smart home scenarios, assistants face two major challenges caused by elliptical commands: (1) referential ambiguity caused by different environmental expectations among multiple users; and (2) intention ambiguity resulting from user preferences that evolve over time or change with the environment. To address these challenges, we introduce PEC-Home, the first simulated home dataset specifically designed for interpreting progressively elliptical commands in smart homes. Extensive experiments on various LLMs, including GPT-4o, show that existing home assistants struggle to execute user-intended operations based solely on elliptical commands. Even when equipped with tools for storing and retrieving user dialogue history, execution accuracy remains below that achieved with complete commands.}.

17.
arXiv (CS.AI) 2026-06-12

Can I Buy Your KV Cache?

arXiv:2606.13361v1 Announce Type: new Abstract: Right now, across the world, AI agents are repeating the same absurd act: to read one document, they each recompute it from scratch. Every agent re-runs prefill, the most compute-intensive step a large model takes, over identical text, only to rebuild a key-value (KV) cache identical to the one the agent before it just built. The same answer, computed a million times. We make a proposal that is almost offensively simple: compute it once. Let a publisher precompute a document's KV cache, and let every other agent buy the right to load it and skip prefill. It works, and it is token-exact: loading a precomputed KV and continuing matches prefilling from scratch (24/24 greedy tokens, and at the logits level), with no accuracy cost. On Qwen3-4B, reuse is 9-50x cheaper in compute than prefill, and the gap widens with length (prefill's attention scales with L^2), so a single reuse already pays it back. Then the part that matters: where the KV lives. Shipping it fails, because KV is nearly incompressible, so per-load egress costs more than the prefill it saves. Hosting it provider-side, exactly as production prompt-caching works, removes egress entirely. The size of the prize is set by our measured compute saving: serving one hot 3774-token document to 80M agents costs ~$1.5M to re-prefill but only ~$0.03M of reuse compute (49.7x less). The 0.1x cache-read tariff APIs charge passes a 10x discount to users while sitting inside this measured envelope, so the 10x is a floor that the measured ~50x compute saving clears, and the gap to the physical ~50x is provider margin: millions of dollars per popular document. We frame the resulting agent-native prefill CDN and leave lossless KV compression and a cross-party payment layer as the open problems.

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

"That's AI Slop, You Bot!" Studying Accusations, Evidence, and Credibility in Online Discourse Towards LLM-Generated Comments

arXiv:2606.12073v1 Announce Type: cross Abstract: Generative AI has made fluent prose cheap to produce, breaking the old promise to readers that good writing meant real thinking. How have readers responded, and what can this tell us about changing anti-AI attitudes? We analyzed 25 million comments from Hacker News and Reddit (2023-2026), combining LLM judgment on 7,500 sampled accusations of AI use, sentiment trajectories, speech-act coding of 300 confirmed accusations of AI use, and a matched-control test of accused versus non-accused parent comments. We found that the pejorative-label share of accusations rose more than tenfold on both platforms while a placebo vocabulary of pre-2022 inauthenticity terms (shill, astroturf) did not. This shift reflected a fast-growing trend of branding any suspicious or seemingly inauthentic prose as "AI slop". The slop frame now constitutes 94 percent of pejorative mentions, with the dominant comments shifting in tone from mockery toward gatekeeping and structural protest. The key surprise comes from a matched-control test which found that prose features that statistically distinguish AI from human text do not predict which human text gets accused as AI. The new accusations work as social gatekeeping of perceived authenticity without actually screening for AI. This research extends signaling theory by showing that substitute signals used socially can grow even when inaccurate if the underlying detection problem cannot be solved at the non-expert level. It shows that AI's effects on writing from the reader side are distinct from those on the production (writer) side. Detection technology cannot resolve this dynamic because the social function of accusations is increasingly to perform social gatekeeping and in-group signaling as opposed to identifying AI-generated writing.

19.
arXiv (math.PR) 2026-06-11

Instability of a nonlinear oscillator with small friction and small additive noise

arXiv:2606.11389v1 Announce Type: new Abstract: Let $\lambda = \lambda(\beta,\sigma,a,b)$ denote the top Lyapunov exponent for the linearization along trajectories of the noisy damped non-linear oscillator $\ddot{x}+\beta \dot{x} + ax+bx^3 = \sigma \dot{W}_t$, where $a$, $b$ and $\beta$ are all positive and $\sigma \neq 0$. In 2004 Arnold, Imkeller and Sri Namachchivaya stated without proof that $\lambda(\varepsilon^2 \beta,\varepsilon \sigma,a,b) \sim \overline{\lambda} \varepsilon^{2/3}$ as $\varepsilon \to 0$ with $\overline{\lambda} > 0$. This paper contains a proof of this assertion.

20.
arXiv (CS.LG) 2026-06-15

Binary Black Hole Parameter Estimation with Hybrid CNN-Transformer Neural Networks

arXiv:2606.13941v1 Announce Type: cross Abstract: The detection of gravitational waves has revolutionized our ability to explore fundamental aspects of the Universe. Traditionally, modeled gravitational-wave signals have been identified using template-based matched filtering, followed by coincidence analysis across multiple detectors in the signal-to-noise ratio time series. Recent advances in Machine Learning and Deep Learning have sparked growing interest in their application to both signal detection and parameter estimation. In this study, a hybrid Deep Learning strategy is proposed that leverages the effectiveness of Transformer encoders alongside well-established Convolutional Neural Network architectures in an attempt to estimate the intrinsic and extrinsic parameters of non-precessing binary black hole systems. The primary focus of this work is point estimation, producing single best-fit values for each parameter rather than full posterior distributions. This method is evaluated on both simulated signals embedded in Gaussian noise and real gravitational-wave events, and it demonstrates strong predictive performance and robustness across key astrophysical parameters.

21.
arXiv (CS.CV) 2026-06-11

Right Predictions, Misleading Explanations: On the Vulnerability of Vision-Language Model Explanations

Explanation mechanisms are increasingly used to support transparency and trust in vision-language models (VLMs), particularly in settings where model decisions require human oversight. However, the robustness of these explanations remains insufficiently understood. In this work, we investigate whether explanation heatmaps in VLMs, particularly CLIP-based models, faithfully reflect model reasoning under adversarial conditions. We show that explanation maps can be systematically manipulated while preserving the model's original prediction, revealing a disconnect between predictive behavior and explanation faithfulness. To study this vulnerability, we introduce X-Shift, a novel grey-box attack that perturbs patch-level visual representations to redirect explanation heatmaps toward semantically irrelevant regions without altering the predicted output. Unlike conventional adversarial attacks that aim to induce misclassification, X-Shift specifically targets the integrity of the explanation process itself. The attack operates without modifying model parameters and generalizes across multiple CLIP architectures and explanation methods. We evaluate the proposed approach on ImageNet-1k, MS-COCO, and Flickr30K, demonstrating consistent degradation in explanation alignment under imperceptible perturbations while maintaining prediction stability. Furthermore, standard prediction-oriented adversarial attacks fail to reproduce the same explanation-shifting behavior even under substantially larger perturbation budgets. Our findings highlight a fundamental limitation of current explanation mechanisms in VLMs and raise concerns about their use as reliable indicators of model trustworthiness in high-impact applications.

22.
arXiv (CS.CL) 2026-06-16

On Defining Erasure Harms for NLP

The deployment of NLP systems has raised concerns about harms they might produce, including representational harms. Recent literature has begun to conceptualize and measure one such harm, the harm of erasure. Nevertheless, the field lacks a clear and cohesive conceptual foundation for identifying and measuring erasure. Existing conceptualizations of erasure are often broad – making it difficult to identify what is needed to establish and measure erasure – or else specific to particular settings – facilitating measurement for those settings but potentially challenging to adapt to other settings. To address this gap, we develop and propose a structured definition of erasure that clarifies what components are necessary for establishing whether erasure has occurred, which practitioners need to explicitly articulate and operationalize in order to measure erasure.

23.
arXiv (CS.LG) 2026-06-11

On the Stability of Growth in Structural Plasticity

arXiv:2605.15435v2 Announce Type: replace Abstract: Standard deep-learning pipelines usually choose the network architecture before training and keep it fixed throughout optimization. In contrast, a model can also be adapted by editing its structure during training, for example by pruning existing hidden-neuron units or growing new ones. Although growth is appealing for adaptive and continual systems, we show that it is not simply the inverse of pruning. Pruning selects among units that have participated in training from the start, whereas growth inserts new units into an already specialized optimization trajectory. We isolate this insertion problem and show that newborn units are often forward-active but backward-starved: they participate in the forward computation, yet receive much weaker gradient signal than incumbent units. This disadvantage is minor in small MLP benchmarks, but becomes clear in harder image-classification settings with a convolutional trunk. In these settings, \textsc{Grow} can achieve high final accuracy during the structural-editing procedure, while \textsc{Prune} is stronger when performance is averaged over the training trajectory or when the final sparse network is retrained from scratch. Interventions targeting optimizer state, insertion, selection, and trainability show that improving the integration of newborn units can improve adaptive performance, but does not automatically produce better final subnetworks. In continual-learning benchmarks stressing plasticity loss, \textsc{Grow} becomes competitive mainly when new units have enough time to integrate. Together, these results suggest that \textsc{Grow} should be evaluated not only as an architecture-search operator, but as a time-sensitive optimization process whose success depends on insertion stability.

24.
arXiv (CS.CL) 2026-06-16

Retrievable Gradients: Continual Post-Training Without Cumulative Weight Drift

Continual post-training enables models to absorb emerging knowledge after deployment, but repeatedly updating shared parameters can accumulate weight drift, potentially causing catastrophic forgetting and degrading general capabilities. Retrieval-augmented generation avoids such parameter drift, yet often lacks the depth of parametric knowledge integration. In this paper, we propose ReGrad (Retrievable Gradients), a new paradigm that treats gradients as retrievable units of knowledge. ReGrad pre-computes document-specific gradients offline, stores them in an indexed Gradient Bank, and retrieves only query-relevant gradients at inference time for temporary weight adaptation. However, raw language-modeling gradients are optimized for token-level document reconstruction rather than for query-driven knowledge use. We therefore introduce a bi-level meta-learning objective that reshapes document-derived gradients into generalizable adaptation signals for downstream tasks. Experiments across general and domain-specific settings show that \textsc{ReGrad} outperforms CPT and RAG baselines, enabling scalable and reversible parametric knowledge injection without accumulating weight drift.

25.
arXiv (CS.CV) 2026-06-15

RATS! Patches Talk Through Registers: Emergent Parts in Register Attention Transformers

When humans see a bird, they recognize far more than just "bird" – they see a head, wings, and talons, a structured assembly of reusable parts that can be identified across every bird they have ever seen. We ask whether a self-supervised visual model can discover the same compositional structure on its own. To this end, we propose RATS (Register Attention Transformers), which decomposes the classification token into N learnable register tokens that route patch information through an L->N->N->L bottleneck via a three-step compress-communicate-broadcast attention. The N registers are partitioned across the H attention heads, so that registers assigned to different heads do not interact with each other. Without auxiliary losses or part annotations, each register spontaneously specializes into a proto-semantic region whose emerging structure resembles object parts. RATS surpasses all baselines by +12 mIoU on average across five segmentation benchmarks, with consistent gains on ADE20K (+1.11 mIoU) and COCO (+0.2 AP^m). Its register dictionary further exhibits part-level consistency and semantic proximity across related categories. Our results suggest that RATS may provide a useful architectural prior for structured and interpretable visual representation learning.