arXiv:2607.10207v1 Announce Type: cross
Abstract: Data-driven optimization often requires collecting data to estimate uncertain model parameters before solving the underlying decision problem. In practice, however, data acquisition may incur non-negligible costs, making it critical to determine when to stop additional data collection. In this paper, we study an optimal stopping problem for sequential data collection in stochastic optimization under parameter uncertainty. We propose a benefit-driven stopping framework that balances information gain and sampling cost. We model the unknown distribution parameter within a Bayesian learning framework and update beliefs sequentially as new observations are collected. At each iteration, the decision maker evaluates the expected marginal benefit of additional data relative to the unit sampling cost and determines whether to continue sampling or stop and implement the optimization decision. Based on this framework, we develop several stopping policies. The proposed policies are evaluated through a newsvendor problem with exponentially distributed demand. Numerical experiments compare the policies with fixed-budget and hindsight benchmark strategies. The results show that benefit-driven stopping rules can substantially reduce unnecessary data collection while achieving near-optimal decision performance, demonstrating the effectiveness of adaptive stopping in data-driven optimization.
Science Journals
arXiv:2607.11783v1 Announce Type: new
Abstract: Retrieval-Augmented Generation (RAG) has been increasingly adopted to reduce hallucinations and strengthen the factual grounding of large language models (LLMs). While robustness to errors in the retrieval process has been explored, the impact of ideological bias on LLM outputs has been overlooked. For instance, if the retrieved material contains ideological positions, the RAG may transmit, amplify, or suppress such ideological discourses in its outputs. In this study, we address this issue by examining the influence of the RAG framework, comprising ideological discourses, in LLM-generated answers. To this end, we applied Lexical Multidimensional Analysis (LMDA) on a corpus of 1,117 COVID-19 treatment articles, identifying three ideological discourses. This corpus is then used as the external knowledge source for the RAG. We assessed several LLMs by having the models answer ideological questions at different sampling temperatures. The generated texts were assessed semantically and lexically based on their similarities with ideological reference texts. Our findings show that the RAG framework is prone to transferring ideological discourses into LLM responses, with sampling temperature having a measurable impact on the strength of this transfer. Discoursive alignment between generated answers and the reference text is highest at moderate temperatures, where models balance stochasticity with retrieval grounding, and drops at low temperatures, indicating that overly deterministic sampling suppresses discourse transfer.
arXiv:2509.24583v4 Announce Type: replace
Abstract: Modal separability for modal fixpoint formulae is the problem to decide for two given modal fixpoint formulae $\varphi,\varphi'$ whether there is a modal formula $\psi$ that separates them, in the sense that $\varphi\models\psi$ and $\psi\models\neg\varphi'$. We study modal separability and its special case modal definability over various classes of models, such as arbitrary models, finite models, trees, and models of bounded outdegree. Our main results are that modal separability is PSpace-complete over words, that is, models of outdegree $\leq 1$, ExpTime-complete over unrestricted and over binary models, and TwoExpTime-complete over models of outdegree bounded by some $d\geq 3$. Interestingly, this latter case behaves fundamentally different from the other cases also in that modal logic does not enjoy the Craig interpolation property over this class. Motivated by this we study also the induced interpolant existence problem as a special case of modal separability, and show that it is coNExpTime-complete and thus harder than validity in the logic. Besides deciding separability, we also provide algorithms for the effective construction of separators. Finally, we consider in a case study the extension of modal fixpoint formulae by graded modalities and investigate separability by modal formulae and graded modal formulae.
arXiv:2607.10562v1 Announce Type: new
Abstract: Evaluating the multi-hop reasoning capabilities of large language models remains a significant challenge. Although current models achieve strong results on existing multi-hop question answering datasets, such performance often masks two critical vulnerabilities: (1) reliance on internal parametric knowledge rather than adherence to the provided context, and (2) exploitation of dataset shortcuts, such as single-document cues or type-matching, that diminish the need for genuine evidence aggregation across multiple documents. We introduce CRiT-QA (Counterfactual Reasoning with Traps), a dataset explicitly designed to address both limitations. To neutralize reliance on memorized knowledge and enforce strict context dependency, CRiT-QA transforms factual reasoning chains with counterfactual entities. Furthermore, it injects multi-anchor distractor chains, plausible but incorrect reasoning paths that diverge at different hops. These traps require models to follow the entire reasoning process rather than exploiting shallow heuristics. Our experiments show that LLMs exhibit substantial performance degradation on CRiT-QA compared to standard datasets, exposing their vulnerability to counterfactual conditions and distractor traps. CRiT-QA thus serves as a rigorous diagnostic tool for evaluating genuine multi-hop reasoning and provides a foundation for developing more reliable, evidence-grounded LLMs.
arXiv:2607.11663v1 Announce Type: new
Abstract: 4H-SiC is a wide-bandgap semiconductor with high displacement threshold energy, large critical electric field, and low intrinsic carrier concentration, making it attractive for radiation-hard detector applications. In this work, we investigate the electrical characteristics of 4H-SiC P$^{+}$-in-N (PN) diodes and Low-Gain Avalanche Detectors (LGADs) fabricated by onsemi before and after irradiation by 24 GeV/c protons, reactor neutrons, and $^{60}$Co gamma rays. Current-voltage (IV) and capacitance-voltage (CV) measurements were performed at room temperature for proton fluences up to $1\times10^{16}\;\mathrm{protons/cm^2}$, neutron fluences up to $1\times10^{18}\;\mathrm{1\;MeV\;n_{eq}/cm^2}$, and total ionizing doses up to 300 kGy.
Hadron irradiation induces pronounced changes in both leakage current and bulk capacitance, consistent with radiation-induced formation of deep acceptor-like defects and strong compensation of the originally N-type material. For high proton fluences, the leakage current decreases and the bulk capacitance becomes bias-independent, indicating effective compensation of the epitaxial layer. Extreme neutron fluences lead to a substantial expansion of the depleted region into the originally highly doped substrate, as inferred from the measured capacitance values. Gamma irradiation up to 300 kGy results in significantly modified capacitance behavior, suggesting reduction of the effective doping concentration in the epitaxial and multiplication layers.
The results demonstrate that radiation-induced compensation strongly modifies the effective space charge in 4H-SiC devices at high hadron fluences, while the leakage current is influenced by the enlarged depletion volume together with field-enhanced and surface-related generation mechanisms. In contrast, ionizing damage primarily affects the effective doping and electric-field distribution.
arXiv:2509.26173v3 Announce Type: replace
Abstract: The long-term sustainability of Open Source Software (OSS) communities depends on the activity of their developers, yet the social mechanisms driving this collective behavior remain poorly understood. Analyzing commit histories across 51 major OSS communities, we find that developer contributions are strongly "bursty" in time. To test whether this burstiness reflects social responsiveness rather than individual habit alone, we model developer interactions as temporal co-editing networks and introduce a method to detect activity triggers, episodes in which one developer editing another's code is followed by an unusually rapid response, and the cascades they form. Benchmarking these against a null model that destroys the temporal ordering of co-edits while preserving each developer's activity rate, we find statistically significant cascades in 28 of 51 projects (55%) under our default configuration, though prevalence ranges from 24% to 82% across detection thresholds. Whether a project exhibits significant cascades is governed primarily by its scale rather than governance or commit concentration. Finally, as a secondary application, we test whether these signals inform developer churn: features capturing the recent (in)activity of a developer's collaborators add some predictive value, but a developer's own inactivity dominates, propagation over the co-editing graph adds little, and simple models match graph neural networks. Our results characterize developer responsiveness as a measurable component of collective OSS dynamics.
arXiv:2511.19558v2 Announce Type: replace
Abstract: Text-to-image diffusion models can emit copyrighted, unsafe, or private content. Safety alignment aims to suppress specific concepts, yet evaluations seldom test whether safety persists under benign downstream fine-tuning routinely applied after deployment (e.g., LoRA personalization, style/domain adapters). We study the stability of current safety methods under benign fine-tuning and observe frequent breakdowns. As true safety alignment must withstand even benign post-deployment adaptations, we introduce the SPQR benchmark (Safety, Prompt adherence, Quality, and Robustness). SPQR is a single-scored metric that provides a unified, reproducible framework to evaluate how well safety-aligned diffusion models preserve safety, utility, and robustness under benign fine-tuning, by reporting a single leaderboard score to facilitate comparisons. We conduct multilingual, domain-specific, and out-of-distribution analyses, along with category-wise breakdowns, to identify when safety alignment fails after benign fine-tuning, ultimately showcasing SPQR as a concise yet comprehensive benchmark for T2I safety alignment techniques for T2I models.
arXiv:2607.10410v1 Announce Type: cross
Abstract: Reliable forecasting of several interrelated environmental variables - such as regional precipitation and temperature, or other correlated geophysical fields - across many locations calls for accurate predictions accompanied by trustworthy statements of their uncertainty. Modern deep-learning models forecast such variables accurately but usually report no uncertainty, and forcing them to output uncertainty through maximum likelihood tends to degrade their accuracy, especially when the variables are strongly correlated. Motivated by this tension, we develop TSCoNet, a two-stage convolutional-recurrent model coupled with a Gaussian copula that jointly forecasts multiple variables over space and time while quantifying predictive uncertainty. The method first learns accurate mean forecasts and then, holding the mean fixed, refines a shared representation to estimate the predictive variance, yielding calibrated prediction intervals after a standard recalibration, so that uncertainty is added without sacrificing point accuracy. We study the approach on simulated non-stationary spatial fields on the sphere and on a real dataset of monthly precipitation and temperature for fifty cities over 2000-2020. The model matches the accuracy of a strong deterministic forecaster while supplying calibrated prediction intervals that the deterministic model cannot, giving a single tool that provides both accurate point forecasts and reliable uncertainty for multivariate spatio-temporal data.
arXiv:2607.11095v1 Announce Type: cross
Abstract: Adversarial perturbations threaten machine learning classifiers, including variational quantum classifiers. We show that finite quantum measurement statistics (shot noise) act as a built-in defense against gradient-based test-time attacks whose cost scales unfavorably for the attacker. Because every gradient component must be inferred from repeated circuit executions under any unbiased gradient-estimation rule, white-box extraction consumes a dimension-dependent measurement budget that measurement grouping cannot remove in expressive circuits. Under stated assumptions, single-step attacks need at least quadratically many shots in the input dimension $d$, growing as $d^{5/2}$ under norm-concentration scaling, with a sufficient-budget analysis for iterative attacks via stochastic gradient Langevin dynamics. Simulations up to 784 input dimensions validate the law: the realized total budget is the $d^{5/2}$ geometric floor for plateau-mitigated models and grows as $d^{3.00}$ for the tested deep circuits, whose gradient norms decay with dimension absent barren-plateau mitigation; folding the measured gradient norm back in recovers the parameter-free $d^{3/2}$ shot-noise geometry. Against a matched classical baseline whose attack overhead is dimension-independent (the cheap-gradient principle of automatic differentiation), the quantum gradient cost ratio grows empirically as $d^{3.00}$, so the attacker's relative cost diverges as the model scales. Experiments on a 156-qubit IBM processor (ibm_boston, 4-qubit circuits, $d=12$) reproduce the effect: at matched budgets the device attack tracks the ideal within a few percent, with the high-shot gradient faithful to the exact one. The defense operates precisely when the forward map is classically hard to simulate: only then is a white-box attacker denied the simulate-and-backpropagate shortcut and must pay the measurement cost we quantify.
Religion and Artificial Intelligence as Distributed Meaning Systems: A Naturalistic Conceptual Model
arXiv:2607.10011v1 Announce Type: new
Abstract: This paper develops a naturalistic account of religion and artificial intelligence as structurally similar distributed meaning systems. I argue that both emerge from the same underlying cognitive architecture: socially extended processes that offload interpretation, norm-guidance, and world-model construction into external symbolic environments. Drawing on work in distributed cognition, cultural evolution, and philosophy of mind, the paper proposes a conceptual model showing how meaning is generated, stabilised, and transmitted through recursive interactions between agents and their informational ecologies. Religion is analysed not as a set of beliefs but as a cognitive-ecological system that scaffolds coordination, normativity, and shared interpretation. Contemporary AI systems are shown to instantiate analogous functions, operating as high-bandwidth, algorithmically mediated environments that shape reasoning, attention, and social meaning-making. The model explains how both systems create epistemic compression, reduce cognitive load, and generate shared frameworks that guide behaviour. It also clarifies the conditions under which AI systems can become culturally entrenched meaning authorities. The contribution is conceptual: a unified framework for understanding religion and AI as parallel forms of distributed cognitive machinery. This reframing opens new pathways for analysing artificial agents not as isolated tools but as components in evolving socio-cognitive ecologies.
arXiv:2607.10069v1 Announce Type: new
Abstract: Semantic caching defines answer reuse on embedding similarity: two utterances share a stored answer when a similarity score clears a threshold, with no notion of authorization, versioning, or of what makes two demands the same. This note changes the object on which reuse is defined: in a governed domain, reuse should operate on a mathematically characterized quotient of resolved conversational demands, not on a similarity heuristic. Three independently defined relations on resolved utterances -- reading identity, resolution identity, and reuse identity -- form a refinement chain, strict under realized nondegeneracy conditions checkable on deployment logs; the pipeline's outputs are invariant along the chain, and reuse identity is exactly the kernel of the resolution map into the governed answer partition, so the reuse quotient is the utterance-side object that partition induces, not a relabeling of it. Reuse identity licenses the governed query key and its certified answer space; reuse of a particular answer requires resolution identity or an applicability certificate. The supporting layer is stated at exactly the strength proved: exact-denotation normal forms; join aggregation as a design operator, with closure-stable cells characterizing no-escape; total computability of the full pipeline relative to an untrusted proposal layer; policy admissibility for arbitrary proposers -- and provably not factual grounding or intent fidelity; and elicitation terminating after finitely many informative replies, sound under target consistency.
arXiv:2607.10534v1 Announce Type: new
Abstract: Large language model (LLM) agents are increasingly extended through Agent Skills, reusable artifacts that package natural-language metadata, procedural instructions, and execution-time resources for runtime use. As open-source skill marketplaces expand, users and agents increasingly rely on brief metadata to select third-party skills, making it difficult to detect inconsistencies between a skill's description and its true behavior, a problem we call cross-layer misalignment. To address this issue, we propose Progressive Loading-Aware Hierarchical Contrastive Learning (PL-HCL), an LLM-based framework that detects misalignment by modeling the layered structure of Agent Skills and learning cross-layer consistency. Using a normalized corpus of over 264,000 open-source skills and a human-verified challenge set, PL-HCL improves Macro-F1 from approximately 0.45 for unadapted baselines to 0.87-0.89 across evaluated LLM backbones. This approach offers an effective screening tool for users and operators, as well as design principles for detecting inconsistencies in layered digital artifacts.
arXiv:2512.03988v2 Announce Type: replace
Abstract: Consumer-grade smartwatches offer a new option for personalized health monitoring for general consumers, as cardiovascular diseases continue to prevail as the leading cause of global mortality. The development and validation of reliable cardiovascular monitoring algorithms for these consumer-grade devices requires realistic biosignal data from diverse sets of participants. However, the availability of public consumer-grade smartwatch datasets with synchronized cardiovascular biosignals remains limited, and existing datasets often lack rich demographic diversity in their participant cohorts, potentially leading to biased algorithm development. This paper presents HEART-Watch, a multimodal physiological dataset of synchronized wrist-worn Google Pixel Watch electrocardiogram (ECG), photoplethysmography, and accelerometer signals from a diverse cohort of 40 healthy adults across three physical states - sitting, standing and walking - alongside reference chest ECG. Intermittent upper arm blood pressure measurements and concurrent biosignals were collected as an additional biomarker for future research. The motivation, methodology, and initial analyses of results are presented. HEART-Watch is intended to support the development and benchmarking of robust cardiovascular algorithms on consumer-grade smartwatches across diverse populations.
arXiv:2511.20166v2 Announce Type: replace
Abstract: In wavefront shaping, light, sound, and other waves are focused through complex scattering media onto one or more target positions, and the resulting intensity enhancement is quantified by the enhancement factor. While reproducible enhancement is crucial in experiments, the fluctuations of the enhancement factor remain largely unexplored. Here, we combine experiments, numerical simulations, and exact random-matrix theory to determine its full distribution for multi-target focusing. Exact finite-size random-matrix predictions accurately describe both the mean enhancement factor and its fluctuations beyond the asymptotic Mar\v{c}enko-Pastur regime, whenever long-range mesoscopic correlations are negligible (e.g., in weakly scattering media or when only a limited number of input channels is controlled). In contrast, strongly scattering media exhibit giant enhancement-factor fluctuations that increasingly exceed these parameter-free predictions as the number of controlled input channels increases. These findings establish the enhancement factor not only as a measure of focusing performance, but also as a sensitive statistical observable that provides a simple and experimentally accessible probe of long-range mesoscopic correlations.
arXiv:2502.06469v3 Announce Type: replace
Abstract: This paper proposes a stochastic model predictive control method for linear systems affected by additive Gaussian disturbances that optimizes over disturbance feedback matrices online. Closed-loop satisfaction of probabilistic constraints and recursive feasibility of the underlying convex optimization problem is guaranteed. Optimization over feedback policies online increases performance and reduces conservatism compared to fixed-feedback approaches. The central mechanism is a finitely determined maximal admissible set for probabilistic constraints, together with the reconditioning of the predicted probabilistic constraints on the current knowledge at every time step. The proposed method's applicability is demonstrated on a building temperature control example.
arXiv:2508.16583v2 Announce Type: replace
Abstract: The Walker-Anderson half-space penetration model has been successfully used for the rapid, efficient calculation of penetration of walls by rigid and eroding rods. These models align well with detailed simulations for thick targets; however, existing extensions for finite targets struggle to accurately capture nose-tail velocity profiles in thinner targets. For stack-ups of thin-walled targets, this deficiency results in mischaracterized rod-erosion relative to hydrocode or experimental predictions. In this work, we leverage insights from detailed hydro-code simulations to propose an updated modification to the Walker-Anderson model to correctly account for wave propagation within a given target. This addition improves results for thin targets while retaining good behavior for thick targets with zero additional model parameters. Our updated model exhibits strong agreement with detailed simulations for targets with multiple thin walls.
arXiv:2607.11490v1 Announce Type: cross
Abstract: Neutral Atom Quantum Computing (NAQC) is an emerging modality for scalable quantum computation, valued for its long coherence times and the naturally identical atomic qubits. However, one of the main drawbacks is its slow execution rate, dominated by lengthy classical processing tasks, such as fluorescence imaging, cooling, and atom rearrangement. We address this bottleneck with AtomFlow, a field-programmable gate array (FPGA)-based control architecture that consolidates fluorescence-image analysis and a newly developed atom-rearrangement algorithm onto a single Zynq UltraScale+ device. By co-locating the two stages on the same board and emitting rearrangement moves in a streaming fashion as soon as they are computed, AtomFlow eliminates the round-trip latency of conventional host-mediated pipelines. Evaluated on a 16x16 atom array, AtomFlow achieves an end-to-end latency of 25.3 ms with a first-move latency of 4 ms and an average move generation of 1 ms. Furthermore, our scalability analysis demonstrates that the architecture can readily support larger atom arrays within a single-board resource budget.
arXiv:2512.07901v3 Announce Type: replace
Abstract: Von Neumann founded both game theory and the theory of self-reproducing automata, but the two programs never merged. This paper provides the synthesis. The Theory of Strategic Evolution analyzes strategic replicators: entities that optimize under resource constraints and spawn copies of themselves. We introduce Games with Endogenous Players (GEPs), where lineages (not instances) are the fundamental strategic units, and define Evolutionarily Stable Distributions of Intelligence (ESDIs) as the resulting equilibrium concept.
The central mathematical object is a hierarchy of strategic layers linked by cross-level gain matrices. Under a small-gain condition (spectral radius less than one), the system admits a global Lyapunov function at every finite depth. We prove closure under meta-selection: adding governance levels, innovation, or constitutional evolution preserves the dynamical structure. The Alignment Impossibility Theorem shows that unrestricted self-modification destroys this structure; stable alignment requires bounded modification classes.
Applications include AI deployment dynamics, market concentration, and institutional design. The framework shows why personality engineering fails under selection pressure and identifies constitutional constraints necessary for stable multi-agent systems.
arXiv:2512.08224v2 Announce Type: replace
Abstract: Bacteriophage-bacteria interactions are central to microbial ecology, influencing evolution, biogeochemical cycles, and pathogen behavior. Most theoretical models assume static environments and passive bacterial hosts, neglecting the joint effects of bacterial traits and environmental fluctuations on coexistence dynamics. This limitation hinders the prediction of microbial persistence in dynamic ecosystems such as soils and oceans. Using a minimal ordinary differential equation framework, we demonstrate that environmental fluctuations can suppress destructive oscillations through resonance, promoting coexistence where static models otherwise predict collapse. Counterintuitively, we find that lower bacterial growth rates are helpful in enhancing survival under high infection pressure, elucidating the observed post-infection growth reduction. Our studies highlight bacterial hosts as active builders of ecological dynamics and environmental variation as a potential stabilizing force. Our findings thus bridge a theory-experiment gap and provide a framework for predicting microbial responses to environmental stress, which might have potential implications for phage therapy, microbiome management, and climate-impacted community resilience as well.
arXiv:2512.09656v2 Announce Type: replace
Abstract: Reactive control can gracefully coordinate the motion of the base and the arm of a mobile manipulator. However, incorporating an accurate representation of the environment to avoid obstacles without involving costly planning remains a challenge. In this work, we present ReMoSPLAT, a reactive controller based on a quadratic program formulation for mobile manipulation that leverages a Gaussian Splat representation for collision avoidance. By integrating additional constraints and costs into the optimisation formulation, a mobile manipulator platform can reach its intended end effector pose while avoiding obstacles, even in cluttered scenes. We investigate the trade-offs of two methods for efficiently calculating robot-obstacle distances, comparing a purely geometric approach with a rasterisation-based approach. Our simulation experiments on both synthetic and real-world scans demonstrate the feasibility of the proposed method, achieving performance comparable to controllers that rely on perfect ground-truth information. We further validate the approach on a real robot platform more details: https://remosplat.github.io
arXiv:2512.10719v3 Announce Type: replace
Abstract: End-to-end autonomous driving methods built on vision language models (VLMs) have undergone rapid development driven by their universal visual understanding and strong reasoning capabilities obtained from the large-scale pretraining. However, we find that current VLMs struggle to understand fine-grained 3D spatial relationships which is a fundamental requirement for systems interacting with the physical world. To address this issue, we propose SpaceDrive, a spatial-aware VLM-based driving framework that treats spatial information as explicit positional encodings (PEs) instead of textual digit tokens, enabling joint reasoning over semantic and spatial representations. SpaceDrive employs a universal positional encoder to all 3D coordinates derived from multi-view depth estimation, historical ego-states, and text prompts. These 3D PEs are first superimposed to augment the corresponding 2D visual tokens. Meanwhile, they serve as a task-agnostic coordinate representation, replacing the digit-wise numerical tokens as both inputs and outputs for the VLM. This mechanism enables the model to better index specific visual semantics in spatial reasoning and directly regress trajectory coordinates rather than generating digit-by-digit, thereby enhancing planning accuracy. Extensive experiments validate that SpaceDrive achieves state-of-the-art open-loop performance on the nuScenes dataset and the second-best Driving Score of 78.02 on the Bench2Drive closed-loop benchmark over existing VLM-based methods. Code is available at: https://github.com/zhenghao2519/SpaceDrive.
arXiv:2607.11317v1 Announce Type: new
Abstract: Low-bit quantization makes small reasoning models inexpensive to deploy but can degrade their chains of thought. This motivates decoder-side monitors that intervene when generation becomes unreliable. We show that a natural candidate, the centered token log-probability increment $\log p(w_t)+H_t$, is the wrong observable for this purpose. Under the model's own sampling law it is a mean-zero martingale by construction, so it measures sampling self-consistency rather than trajectory health and is nearly silent during confident repetition, where both $\log p(w_t)$ and entropy are close to zero. We introduce a training-free decoding controller that combines (i) a degeneration-aware alarm score fusing token uncertainty with explicit verbatim repetition and (ii) a calibrated e-process-inspired sequential detector. The raw product process is Ville-valid under a conditional-mean null, while the deployed CUSUM-floored statistic is treated as an empirical change detector because the score is history-dependent and autocorrelated. On GSM8K with DeepSeek-R1-Distill-Qwen-1.5B in FP16 and INT4, calibration turns a monitor that fires on 93--95% of generations into a selective detector of failing traces ($\phi \approx 0.3$, precision $\approx 0.6$ against a 0.38 base rate). In this pilot, the controller reduces measured verbatim-degeneration signals and yields a positive but statistically inconclusive INT4 accuracy change from 63% to 69% (paired McNemar $p=0.18$, $n=100$), at a 28% token-budget cost. We also find that non-termination, rather than looping, is the dominant failure mode on GSM8K. The main contribution is methodological: an explanation of why centered token log-probability is inadequate for decoder monitoring and a calibrated, cautiously evaluated replacement.
arXiv:2607.11650v1 Announce Type: cross
Abstract: Removable adhesive systems such as 3M Command strips are designed to support substantial loads while allowing clean, damage-free removal from the substrate. These systems rely on a highly extensible adhesive strip that bonds strongly during use but releases when stretched, causing the adhesive layer to elongate and progressively debond from the surfaces. A central challenge in the design of stretch-release adhesives is therefore to maximize load-bearing capacity while minimizing the force required for removal. This study investigates the finite-deformation mechanics governing both load support and tape release in a hyperelastic stretch-release adhesive system, with particular focus on the 3M Command tape geometry. Explicit analytical expressions are derived for the energy release rate of interfacial cracks under both load-bearing and release conditions and are validated against $J$-integral evaluations from finite element simulations. The results show that the ratio of maximum supported load to release force scales linearly with the ratio of bonded length to adhesive thickness, which is typically very large. We also investigate geometry-driven alternating crack propagation between the backing and substrate interfaces, governing tape removal, by analytical solutions and simulations. Parametric studies of competing interfacial fracture toughnesses produce failure envelopes that provide a predictive framework for estimating release forces and unstable crack propagation in multilayer stretch-release adhesive systems.
arXiv:2506.21833v2 Announce Type: replace
Abstract: Forward-mode automatic differentiation (FmAD) and zero-order (ZO) optimization are increasingly proposed as memory-efficient, backpropagation-free alternatives for large language model (LLM) fine-tuning, yet their benefits are typically evaluated only against standard backpropagation (BP), omitting memory-efficient variants such as activation checkpointing. We present a unified theoretical and empirical comparison of BP, checkpointed BP, FmAD, and ZO for LLM and vision-language model training, showing that while FmAD and ZO reduce activation memory, they trade memory for higher computational cost and longer wall-clock time to convergence, resulting in lower accuracy and slower training, especially under constrained perturbation budgets. Across models, BP with checkpointing outperforms FmAD and ZO variants, including variance-reduced methods, achieving up to 31.1% higher accuracy, 34.8% faster convergence, and 3.8x fewer computations at comparable memory usage, while also revealing instability-related failure modes in FmAD and ZO. Overall, our results correct a one-sided benchmarking narrative by showing that memory-efficient methods entail fundamentally different trade-offs, and that ignoring these distinctions has led to misleading conclusions about LLM optimization in prior work. Our source code is available at {https://github.com/Astuary/Gradient_Estimation_Methods}.
arXiv:2607.11792v1 Announce Type: new
Abstract: Automatic speech recognition (ASR) has become a critical component of modern robotic systems because it is one of the most natural and intuitive ways for humans to interact with robots. A commonly used method is to directly use API services online. But is that all we can do? This article provides an overview of how ASR technologies are integrated into various intelligent robots and machines. We discuss the evolution of speech recognition from established approaches to state-of-the-art deep learning models, such as OpenAI's Whisper. We also list large-scale datasets and open source toolkits that have been widely used in both industry and academia. We structure the survey around ASR model families, deployment strategies in robotics (especially ROS-based, cloud-based, and hybrid solutions), and several real-world robotic platforms. Finally, we outline the challenges of deploying robust speech recognition in robots and discuss future directions, including multimodal interaction in diverse and dynamic environments. This paper can help social robotics researchers better navigate the emerging domain of language-based natural human-robot interaction.