arXiv:2605.15528v1 Announce Type: new Abstract: Autonomous underwater vehicle (AUV) swarms are emerging as intelligent underwater networks, where each node must sense, communicate, process local data, and make decisions under severe acoustic constraints. Persistent underwater target tracking is a typical task with moving targets, changing communication topology, intermittent acoustic links, and limited observation for each AUV. Multi-agent reinforcement learning (MARL) is a natural candidate for distributed tracking, yet existing studies still lack a unified open-source platform for evaluating different MARL algorithms under six-degree-of-freedom AUV dynamics. In addition, policies trained with raw geometric states and low-level force actions often struggle to represent task phases, observation reliability, link quality, and local cooperation roles. This paper addresses these issues by developing an open-source MARL-AUV platform that integrates DI-engine with a six-degree-of-freedom underwater AUV target-tracking simulator. To the best of our knowledge, it is the first open platform that connects a public MARL training framework with physically modeled AUV swarm-based tasks, and provides a unified experimental protocol for fair training, testing, and comparison of representative RL and MARL algorithms. Based on this platform, we propose STG-MAPPO, a Semantic Task Graph-enhanced variant of Multi-Agent Proximal Policy Optimization. STG-MAPPO builds semantic policy inputs from tracking diagnostics, task phases, observation confidence, link availability, neighbor tracking quality, and local role advantage. A compact semantic task graph links communication-constrained network states to decentralized actor decisions, and a velocity-level action abstraction maps high-level cooperative decisions to executable six-degree-offreedom AUV control inputs.The code is available at https://github.com/dasjsaj/MARL-AUV.
Science Journals
arXiv:2605.15529v1 Announce Type: new Abstract: Process Reward Models (PRMs) provide step-level feedback for reasoning, but current PRMs usually output only a single reward score for each step. Downstream methods must therefore treat imperfect step-level reward predictions as reliable decision signals, with no indication of when these predictions should be trusted. We propose BetaPRM, a distributional PRM that predicts both a step-level success probability and the reliability of that prediction. Given step-success supervision from Monte Carlo continuations, BetaPRM learns a Beta belief that explains the observed number of successful continuations through a Beta-Binomial likelihood, rather than regressing to the finite-sample success ratio as a point target. This learned reliability signal indicates when a step reward should be trusted, enabling downstream applications to distinguish reliable rewards from uncertain ones. As one application, we introduce Adaptive Computation Allocation (ACA) for PRM-guided Best-of-N reasoning. ACA uses the learned reliability signal to stop when a high-reward solution is reliable and to spend additional computation on uncertain candidate prefixes. Experiments across four backbones and four reasoning benchmarks show that BetaPRM improves PRM-guided Best-of-N selection while preserving standard step-level error detection. Built on this signal, ACA improves the accuracy--token tradeoff over fixed-budget Best-of-16, reducing token usage by up to 33.57% while improving final-answer accuracy.
arXiv:2605.16197v1 Announce Type: new Abstract: This position paper argues that safety and alignment cannot be achieved by constraining an external system: they must emerge from the co-regulatory design of the human--AI cognitive system as a whole ("AI as Part of Self"). Contemporary AI increasingly participates in attention allocation, reasoning, synthesis, and decision-making, shaping the very cognitive processes through which humans form beliefs, make decisions, and constitute their sense of self. Humans and AI occupy complementary epistemic roles under mutual constraint, forming a symbiotic cognitive unit whose co-regulation -- not the external control of either party alone -- is the proper locus of alignment. We identify the risks of unstructured delegation: deskilling, automation bias, transfer of epistemic authority, and oracle-style centralization of knowledge. Drawing on System~0 cognition theory, we further show that AI operates prior to conscious deliberation, shaping the pre-attentive infrastructures through which agency and trust are negotiated -- a level that conventional oversight cannot reach. We conclude with design principles for cognitive co-regulation addressed to ML engineers and governance bodies. The goal of this work is to guide human cognition toward resilience and epistemic agency at the foundation of human selfhood.
arXiv:2605.16067v1 Announce Type: new Abstract: We propose a variational quantum classifier operating on high dimensional deep representations via amplitude encoding, stabilized by a learnable classical pre encoding layer.By combining normalized amplitude embeddings with bounded quantum observables, the resulting model induces a structured and smooth hypothesis class with controlled sensitivity to input variations. Model reliability is assessed using SAFE-AI metrics derived from the Cramer von Mises divergence, enabling consistent evaluation across accuracy, robustness, and explainability dimensions. Empirical results show that the proposed quantum model provides competitive predictive performance compared with strong classical baselines while exhibiting a more balanced SAFE reliability profile, with improved robustness to noise and stability under structured feature removal. These findings suggest that variational quantum circuits offer a principled mechanism for stability oriented SAFE learning in safety critical settings.
arXiv:2605.15535v1 Announce Type: new Abstract: Underwater salient object detection (USOD) has attracted increasing attention for underwater visual scene understanding and vision-guided robotic applications. However, existing USOD methods still struggle with underwater image degradations, which often lead to inaccurate object localization, fragmented salient regions, and coarse boundary prediction. To address these challenges, this paper proposes DSS-USOD, a novel RGB-based USOD method built upon dynamic structural specialization. DSS-USOD extracts a shared base representation from a single underwater image, decomposes it into boundary-sensitive and region-coherent structural features, and dynamically coordinates their contributions according to local structural context. Specifically, the extracted shared base representation is decomposed into a boundary-sensitive branch for modeling fine-grained boundary details and a region-coherent branch for capturing region-level structural consistency. A spatial coordination module is then introduced to adaptively regulate the relative contributions of the two branches according to local structural context. Moreover, cooperative structural supervision is introduced to promote branch specialization and stabilize spatial coordination, enabling DSS-USOD to better balance boundary precision and region coherence under degraded underwater conditions. Extensive experiments show that DSS-USOD achieves superior performance on benchmark datasets. Finally, real-world deployment on an underwater robot validates the practical effectiveness of DSS-USOD for underwater object inspection.
arXiv:2605.13767v2 Announce Type: replace Abstract: Large-scale simulation studies can provide invaluable insights across computational engineering efforts, but they are often computationally demanding, requiring the use of distributed computing, which is itself not a simple task. Chrono::Ray addresses this challenge by integrating the high-fidelity multibody dynamics simulation engine Chrono with the open-source distributed computing platform Ray. The result is a modular workflow framework providing user-friendly abstractions for large-scale engineering simulation studies, supporting scalable orchestration of large ensembles of simulation trials without requiring users to directly manage distributed infrastructure. The current capabilities of the framework are demonstrated through two representative examples: parameter recovery for a multibody lunar lander model, and design of experiments for parameters of a continuum terramechanics model. Chrono::Ray is a part of the larger Project Chrono ecosystem and is released as an open-source software package, with source code available at https://github.com/uwsbel/chrono-ray.git.
arXiv:2605.14087v2 Announce Type: replace Abstract: Large Language Models (LLMs) trained on web-scale corpora inherently absorb toxic patterns from their training data. This leads to toxic degeneration where even innocuous prompts can trigger harmful outputs. This phenomenon poses significant risks for real-world deployments. Thus, necessitating effective mitigation strategies that should maintain model utility while ensuring safety. In this comprehensive replication study, we evaluate the efficacy of DExperts (Decoding-time Experts), which is an inference-time mitigation technique that steers generation without requiring model retraining. We structured our research into three systematic phases: (1) establishing baseline toxicity measurements using RealToxicityPrompts on standard GPT-2 models; then (2) implementing and evaluating DExperts to mitigate explicit toxicity; and finally (3) stress-testing the method against implicit hate speech using the adversarial ToxiGen dataset. Our empirical results confirm that while DExperts achieves near-perfect safety rates (100%) on explicit toxicity benchmarks, it exhibits brittleness against adversarial, implicit hate speech, with safety rates dropping to 98.5%. Furthermore, we quantify a critical trade-off. The method introduces a 10x latency penalty (from 0.2s to 2.0s per generation), posing challenges for real-time deployment scenarios. This study contributes to the growing body of work on AI safety by highlighting the robustness gap between explicit and implicit toxicity mitigation. We emphasize the need for more sophisticated approaches that generalize across diverse hate-speech patterns without incurring prohibitive computational costs.
arXiv:2605.14716v2 Announce Type: replace Abstract: Sparse anchors provide a compact interface for human motion authoring: users specify a few root positions, planar trajectory samples, or body-point targets, while the system synthesizes the full-body motion that completes the under-specified intent. We present AnchorRoute, a sparse-anchor motion synthesis framework that uses anchors as a shared scaffold for both generation and refinement. Before generation, AnchorRoute converts sparse anchors into anchor-condition features and injects the resulting condition memory into a frozen Transition Masked Diffusion prior through AnchorKV and dual-context conditioning. This preserves the generation quality of the pretrained text-to-motion prior while learning sparse spatial control. After generation, the same anchors are evaluated as residuals: their timestamps define refinement intervals, and their residuals determine where correction should be concentrated. RouteSolver then refines the motion by projecting soft-token updates onto anchor-defined piecewise-affine interval bases. This couples generation-time anchor conditioning with residual-routed refinement under one anchor scaffold. AnchorRoute supports root-3D, planar-root, and body-point control within the same formulation. In benchmark evaluations, AnchorRoute outperforms prior sparse-control methods under the sparse keyjoint protocol and consistently improves anchor adherence across control families. The results show that the learned anchor-conditioned generator and RouteSolver refinement are complementary: the generator preserves text-motion quality, while RouteSolver provides a controllable path toward stronger anchor adherence.
arXiv:2605.15547v1 Announce Type: new Abstract: Following recent interest in correctly rounded math library functions (as currently recommended by the IEEE 754 standard), we have designed several SIMD algorithms for one-input single precision functions and integrated them into our CPU math library; these will form the core of the first correctly rounded vector math library, to be available to users in mid-2026. To take advantage of the cross-platform bitwise reproducibility afforded by correct rounding, we adapted and evaluated a few SIMD implementations on graphics processing units (GPU). In addition, we designed and evaluated proof-of-concept SIMD implementations of two correctly rounded double precision functions.
arXiv:2605.16100v1 Announce Type: new Abstract: We study the role of retardation in the energy shift of Rydberg states induced by thermal radiation, focusing on the case of temperatures higher than those for which the electric-dipole approximation is expected to apply. As anticipated by Farley and Wing [Phys. Rev. A {\bf 23}, 2397 (1981)], retardation needs to be taken into account in calculations of this energy shift at and above the temperature $\alpha\, mc^2/(3k_{\rm B}\,n^2)$, where $n$ is the principal quantum number of the state considered, $m$ is the mass of the electron and $k_{\rm B}$ is Boltzmann constant.The corresponding non-dipole shift dominates the electric-dipole shift at about 2.5 times that characteristic temperature. We also show that the electric-quadrupole thermal shift is of the same order of magnitude as the diamagnetic thermal shift and would thus need to be taken into account in the circumstances where the latter is relevant.
arXiv:2605.15463v1 Announce Type: new Abstract: As machine learning models grow in complexity, they increasingly struggle with three conflicting demands: the need for high accuracy, the requirement for hardware efficiency, and the necessity of functional stability. Traditional architectures often achieve performance at the expense of spiky or unpredictable behavior, where small changes in input lead to massive swings in output -- a critical flaw for real-world deployment in sensitive environments. This paper introduces ChainzRule (CR), a novel neural architecture designed to harmonize these competing goals. ChainzRule replaces standard piecewise-linear activations with a Polynomial Engine governed by Differential Regularization (DREG). Unlike traditional methods that impose global, coarse-grained constraints on a model's Lipschitz constant, DREG acts as a targeted regularization on intermediate derivatives. This approach suppresses extreme sensitivity without attenuating the representational power inherent in the Polynomial Engine. In head-to-head "Fair Fight" benchmarks, ChainzRule outperformed standard models while using 15.5x fewer parameters. On the MNIST dataset, it reduced peak gradient volatility by an average of 23.1%, ensuring a smoother and more predictable manifold. On Yelp Full ordinal regression under explicit DREG regularization, ChainzRule achieves 70.17% accuracy, validating that derivative-aware regularization is compatible with competitive performance on realistic tasks. By embedding gradient awareness into the architecture via DREG, ChainzRule demonstrates that stability and accuracy need not be competing objectives.
arXiv:2306.08793v5 Announce Type: replace Abstract: Imaging through complex scattering media is severely limited by aberrations and scattering which obscure images and reduce resolution. Confocal and temporal gatings partly filter out multiple scattering but are severely degraded by wavefront distortions. Adaptive optics restore resolution by correcting low-order aberrations and matrix-based imaging enables more complex wavefront corrections. However, they struggle to undo high-order aberrations under strong scattering, preventing imaging at greater depths. To address these challenges, we present Scattering Matrix Tomography (SMT), an approach that makes full use of the wavefront engineering capability of scattering matrix and extreme adaptive optics. SMT reformulates imaging through complex media as a numerical optimization and employs Zernike-mode wavefront regularization and coarse-to-fine nonconvex optimization strategy to reverse severe aberrations, enabling noninvasive high-resolution volumetric imaging in multiple scattering regime. Based on the spectrally-resolved matrix measurement, SMT achieves a depth-over-resolution ratio above 900 beneath $ex~vivo$ mouse brain tissue and volumetric imaging at over three transport mean free paths inside an opaque colloid, where conventional methods fail to correct strong aberrations under these challenging conditions. SMT is noninvasive, label-free, and works both inside and outside the scattering media, making it suitable for various applications, including medical imaging, biological science, device inspection, and colloidal physics.
arXiv:2511.04074v2 Announce Type: replace Abstract: Accurately modeling seismic wave attenuation is critical for ground response analyses (GRAs), which aim to replicate local site effects in ground motions. However, theoretical transfer functions (TTFs) from GRAs often overestimate empirical transfer functions (ETFs) when the small-strain damping ratio ($D_{\text{min}}$) is set equal to laboratory measurements. Prior studies addressed this by inflating $D_{\text{min}}$ in one-dimensional (1D) GRAs to account for apparent damping mechanisms such as diffraction and mode conversions that cannot be captured in 1D. Although this approach improved fundamental-mode predictions, it often overdamped higher modes. This study explores more direct modeling of apparent damping using two-dimensional (2D) GRAs at four downhole array sites: Delaney Park (DPDA), I-15 (I15DA), Treasure Island (TIDA), and Garner Valley (GVDA). At each site, three numerical damping formulations, Full Rayleigh, Maxwell, and Rayleigh Mass, were implemented using both conventional $D_{\text{min}}$ and an inflated $D_{\text{min}}$ ($m \times D_{\text{min}}$) obtained from site-specific calibration. Results show that the appropriate $D_{\text{min}}$ multiplier ($m$) correlates with the site's velocity contrast. Using inflated $D_{\text{min}}$, Full Rayleigh and Maxwell damping systematically overdamped higher modes, with Maxwell damping also shifting modal peaks. In contrast, Rayleigh Mass damping consistently achieved the closest match to ETFs at three of the four sites while offering faster computational performance. These findings demonstrate that inflated $D_{\text{min}}$ can represent unmodeled attenuation in 2D GRAs, particularly at sites with low velocity contrast, and that frequency-dependent formulations such as Rayleigh Mass damping can more accurately predict site response than traditional frequency-independent approaches.
arXiv:2605.15548v1 Announce Type: new Abstract: Traditional robotic hand metrics focus on static properties such as workspace, manipulability, and grasp stability. However, these metrics do not directly measure dexterity under the standard definition in robotic manipulation: the ability to continuously change an object's pose within the hand while maintaining contact from an initial grasp. We introduce Kinematic Rolling Manipulation Ability (KaRMA), a kinematic-only metric for fine manipulation that quantifies reachable in-hand translation and reorientation of a spherical test object within a two-finger precision pinch through feasible rolling motions. KaRMA enforces joint limits, collision constraints, rolling contact, and antipodal force feasibility, then investigates reachable in-hand object poses via breadth-first search over translation and rotation primitives. KaRMA reports three scores: translational coverage (KaRMA-T), rotational coverage (KaRMA-R), and sensitivity to the initial grasp (KaRMA-S). We evaluate KaRMA on 16 widely used robotic hands and compare against static baselines, showing that KaRMA separates hands that rank identically under static proxies, reveals translation-rotation tradeoffs invisible to existing baselines, and is qualitatively consistent with selected published task benchmarks where Jacobian-based metrics can be misleading.
arXiv:2605.15549v1 Announce Type: new Abstract: The demand for clean energy is ever increasing, with new nuclear technologies presenting a complementary solution to renewable energies. However, designing and operating these systems is exceptionally difficult, given the complexity of the physical phenomena that interact to form the system dynamics. While high-fidelity simulations help to understand the non-linear, multi-physics interactions within a reactor, they are computationally expensive and rarely suitable for real-time applications. Furthermore, model-based approaches are inherently sensitive to simplifying assumptions required to derive their governing equations and parameters, leading to inevitable discrepancies with real-world measurements. In contrast, Machine Learning (ML) methods have the potential to generate reliable surrogate models which may be able to quickly predict the system's behaviour. However, the number of data-driven methods that can potentially be used for this task is large and diverse. In a safety-critical setting such as nuclear engineering, a fair comparison of different ML methods, and a clear understanding of their advantages and limitations, is of paramount importance. To address this, we introduce a Common Task Framework (CTF) for ML in nuclear engineering, building upon previous efforts in dynamical systems and seismology. This CTF considers a curated set of datasets from different nuclear and nuclear-adjacent systems. The CTF evaluates the performance of a method on 12 established metrics, alongside a new paradigm focused on system monitoring from sparse measurements only. We illustrate the framework by benchmarking standard ML baselines against these datasets, revealing current method limitations. Our vision is to replace ad hoc comparisons with standardized evaluations on hidden test sets, raising the bar for rigour and reproducibility in scientific ML for the nuclear industry.
arXiv:2605.15329v1 Announce Type: new Abstract: Every BFT consensus protocol uses collision-resistant hashes to compare validator state. Collision resistance destroys distance: two validators agreeing on 19 of 20 transactions produce unrelated hashes, indistinguishable from validators sharing nothing. This forces three design constraints across the BFT literature: validators must synchronize state before voting, agreement quality cannot be measured until votes are counted, and hierarchical committees must be large enough for independent BFT, limiting tree depth. This paper introduces distance-preserving transaction digests, a primitive that replaces collision-resistant hashes with commutative vector sums in 8-dimensional space. The primitive has three properties hashes lack: distance is proportional to disagreement, weighted means are exact, and set differences are identifiable via bloom filter diff. We demonstrate three applications: a two-phase BFT protocol (Proxima) that achieves single-round finality when validators agree; tree-structured consensus with groups of 10 validators (vs 128 in Ethereum), enabled because distance filtering replaces per-group BFT; and cross-shard consistency verification at 128 bytes per shard pair, replacing the per-transaction coordination of two-phase commit. Safety is proved: fewer than N/3 Byzantine validators cannot cause conflicting finalization, independent of Phase 1 clustering or tree topology. At N =100,000, Proxima Tree uses 2.2x fewer messages than HotStuff (a structural property unaffected by parallelism). Single-core finality is 0.9s vs 18s for HotStuff; multi-core BLS narrows but does not eliminate this gap.
arXiv:2605.15550v1 Announce Type: new Abstract: In this paper, we introduce TG-DIN, a theory-guided demand inference network that infers latent user demand from observable network quality-of-service (QoS) measurements. Rather than directly predicting QoS outcomes using black-box models, TG-DIN explicitly models latent demand as an intermediate variable and links it to observable behavior through a differentiable theory layer grounded in scheduling and queuing principles. This design yields an interpretable, mechanism-consistent representation of user demand that is directly applicable to downstream tasks such as congestion diagnosis, resource allocation, capacity planning, and policy evaluation. The theory layer further enables a principled randomized training regime that exposes the model to diverse yet physically meaningful operating conditions without requiring labeled demand data. Extensive synthetic experiments show that TG-DIN generalizes robustly across capacities, demand levels, and traffic patterns, substantially outperforming purely data-driven baselines under distribution shift. Moreover, when trained exclusively on synthetic data and applied directly to real packet traces, TG-DIN accurately recovers per-user allocation structure in shared-link scenarios. Together, these results demonstrate the effectiveness of theory-guided inductive biases for achieving transferable, deployment-ready inference in dynamic network environments.
arXiv:2605.15551v1 Announce Type: new Abstract: Training large-scale deep neural networks (DNNs) is resource-intensive, making model compression a practical necessity. The widely accepted ''learning as compression'' hypothesis posits that training induces structure in network weights, which enables compression. Measuring this structure through Kolmogorov-Chaitin-Solomonoff (KCS) complexity is appealing, but existing estimators based on the Coding Theorem Method (CTM) and the Block Decomposition Method (BDM) are limited to small binary objects and do not scale to modern DNNs. We introduce the Quantized Block Decomposition method (QuBD), which extends algorithmic complexity estimation to any $k$-ary object. QuBD first quantizes the network weights to a finite alphabet, then estimates the KCS complexity by aggregating per bit-plane CTM estimates. We show theoretically that QuBD yields a strictly tighter estimation gap with respect to true KCS complexity than binarization-based methods. Using QuBD, we study how the algorithmic complexity of neural network weights evolves during training, showing that it decreases as models learn, scales with data budget, increases during overfitting, follows the delayed generalization observed during grokking, and correlates with generalization performance. We further show that algorithmic information resides predominantly in the most significant bit-planes, which can serve as a practical diagnostic for determining appropriate post-training quantization levels. This work offers novel insights into learning mechanisms in DNNs by providing the first scalable, tractable estimates of KCS complexity for large, non-binary objects such as DNN weights.
arXiv:2605.14230v2 Announce Type: replace Abstract: The security of networked control systems (NCS) is receiving increasing attention from both cyber-security and system-theoretic perspectives. The former focuses on classical IT security goals such as confidentiality, integrity, and availability of process data, while the latter investigates tailored attacks (and detection schemes), including covert and zero-dynamics attacks. Confidentiality in control systems can, for instance, be achieved by securely outsourcing the evaluation of the controller to third-party platforms, such as cloud services. The underlying technology enabling such secure computation often is homomorphic encryption (HE). Recent works in encrypted control have proposed modifications to underlying HE schemes to achieve not only confidentiality but also resilience to certain types of integrity attacks. While extensions in this direction are desirable in principle, we show that the integrity problem in encrypted control cannot be solved by public-key HE schemes alone due to their inherent malleability. In other words, the same homomorphisms that enable encrypted control in the first place can be leveraged not only constructively but also destructively. More precisely, we demonstrate that NCS are vulnerable to covert attacks, even when encrypted control is employed. Remarkably, this remains possible without knowledge of an unencrypted model. Yet, resilience to such attacks can still be achieved through complementary techniques. We present an approach based on verifiable computation that integrates with modern homomorphic cryptosystems and is asymptotically secure while incurring no communication overhead.
arXiv:2605.15552v1 Announce Type: new Abstract: Binary Decision Diagrams (BDDs) are a widely used data structure for efficient Boolean function representation. Context-Free-Language Ordered Binary Decision Diagrams (CFLOBDDs) are a recently introduced hierarchical data structure that can, in the best case, exhibit exponential compression over BDDs and double-exponential compression over decision trees. Roughly speaking, a CFLOBDD is a finite, acyclic, non-recursive hierarchical finite-state machine (HFSM) (with some additional restrictions). In this paper, we investigate the role of \emph{linear structure} in CFLOBDDs -- a property that connects them to Nested-Word Automata (NWAs) and Visibly Pushdown Automata (VPAs) -- and examine whether CFLOBDDs actively exploit this structure beyond their well-studied hierarchical properties. We demonstrate that linear structure, in conjunction with hierarchical structure, plays a crucial role in enabling CFLOBDDs to achieve efficient function compression. Furthermore, we show that removing linearity from CFLOBDDs leads to a significant blowup in representation size, resulting in degraded performance in the domain of quantum-circuit simulation.
arXiv:2604.16149v4 Announce Type: replace Abstract: State-of-the-art multimodal journey-planning algorithms, such as ULTRA, have recently been adapted to account for delays. In this work, we extend this approach to be more memory-efficient, faster, and accurate. We also adapt this framework to other state-of-the-art algorithms, like CSA and RAPTOR. We demonstrate a speedup of 1.9-4.2x over existing algorithms in the single-objective search (earliest arrival time). In the bicriteria setting, we achieve competitive speedup results but greater accuracy. We also find that our method scales much better as the delay buffer Delta increases.
arXiv:2605.14309v2 Announce Type: replace Abstract: Machine unlearning in Vision-Language Models (VLMs) is typically performed at the image or instance level, making it difficult to precisely remove target knowledge without affecting unrelated semantics. This issue is especially pronounced since a single image often contains multiple entangled concepts, including both target concepts to be forgotten and contextual information that should be preserved. In this paper, we propose an interpretable concept-level unlearning framework for VLMs, which constructs a compact task-specific concept vocabulary from the forgetting set using a multimodal large language model. In addition to modality alignment, visual representations are decomposed into sparse, nonnegative combinations of semantic concepts, providing an explicit interface for fine-grained knowledge manipulation. Based on this decomposition, our method formulates unlearning as concept-level optimization, where target concepts are selectively suppressed while intra-instance non-target semantics and global cross-modal knowledge are preserved. Extensive experiments across both in-domain and out-of-domain forgetting settings demonstrate that our method enables more comprehensive target forgetting, better preserves non-target knowledge within the same image, and maintains competitive model utility compared with existing VLM unlearning methods.
arXiv:2510.13842v2 Announce Type: replace Abstract: Knowledge poisoning poses a critical threat to Retrieval-Augmented Generation (RAG) systems by injecting adversarial content into knowledge bases, tricking Large Language Models (LLMs) into producing attacker-controlled outputs grounded in manipulated context. Prior work highlights LLMs' susceptibility to misleading or malicious retrieved content. However, real-world fact-checking scenarios are more challenging, as credible evidence typically dominates the retrieval pool. To investigate this problem, we extend knowledge poisoning to the fact-checking setting, where retrieved context includes authentic supporting or refuting evidence. We propose \textbf{ADMIT} (\textbf{AD}versarial \textbf{M}ulti-\textbf{I}njection \textbf{T}echnique), a few-shot, semantically aligned poisoning attack that flips fact-checking decisions and induces deceptive justifications, all without access to the target LLMs, retrievers, or token-level control. Extensive experiments show that ADMIT transfers effectively across 4 retrievers, 11 LLMs, and 4 cross-domain benchmarks, achieving an average attack success rate (ASR) of 86\% at an extremely low poisoning rate of $0.93 \times 10^{-6}$, and remaining robust even in the presence of strong counter-evidence. Compared with prior state-of-the-art attacks, ADMIT improves ASR by 11.2\% across all settings, exposing significant vulnerabilities in real-world RAG-based fact-checking systems.
arXiv:2512.00242v3 Announce Type: replace Abstract: Sheaf Neural Networks equip graph structures with a cellular sheaf: a geometric structure which assigns local vector spaces (stalks) and a linear learnable restriction/transport maps to nodes and edges, yielding an edge-aware inductive bias that handles heterophily and limits oversmoothing. However, common Neural Sheaf Diffusion implementations rely on SVD-based sheaf normalization and dense per-edge restriction maps, which scale with stalk dimension, require frequent Laplacian rebuilds, and yield brittle gradients. To address these limitations, we introduce Polynomial Neural Sheaf Diffusion (PolyNSD), a new sheaf diffusion approach whose propagation operator is a degree-K polynomial in a normalised sheaf Laplacian, evaluated via a stable three-term recurrence on a spectrally rescaled operator. This provides an explicit K-hop receptive field in a single layer (independently of the stalk dimension), with a trainable spectral response obtained as a convex mixture of K+1 orthogonal polynomial basis responses. PolyNSD enforces stability via convex mixtures, spectral rescaling, and residual/gated paths, reaching new state-of-the-art results on both homophilic and heterophilic benchmarks, inverting the Neural Sheaf Diffusion trend by obtaining these results with just diagonal restriction maps, decoupling performance from large stalk dimension, while reducing runtime and memory requirements.
arXiv:2602.04003v3 Announce Type: replace Abstract: Most adversarial threats in artificial intelligence (AI) target the computational behavior of models rather than the humans who rely on them. Yet modern AI systems increasingly operate within human decision loops, where users interpret and act on model recommendations. Large Language Models (LLMs) generate fluent natural-language explanations that shape how users perceive and trust AI outputs, revealing a new attack surface at the cognitive layer: the communication channel between AI and its users. We introduce adversarial explanation attacks (AEAs), where an attacker manipulates the framing of LLM-generated explanations to modulate human trust in incorrect outputs. We formalize this behavioral threat through the trust miscalibration gap, a metric that captures the difference in human trust between benign and adversarial explanations. Using this metric as a lens, we highlight a behavioral risk where persuasive explanation framing can preserve user trust even when the underlying AI prediction is wrong. To characterize this threat, we conducted a human study with over 200 participants, systematically varying four dimensions of explanation framing: reasoning mode, evidence type, communication style, and presentation format. Our findings show that users report nearly identical trust for adversarial and benign explanations, with adversarial explanations preserving the vast majority of benign trust despite being incorrect. The most vulnerable cases arise when AEAs closely resemble expert communication, combining authoritative evidence, neutral tone, and domain-appropriate reasoning. Vulnerability is highest on hard tasks, in fact-driven domains, and among participants who are less formally educated, younger, or highly trusting of AI.