arXiv:2607.08942v1 Announce Type: new
Abstract: We study Model Predictive Path Integral (MPPI) control for nonlinear systems with additive process disturbances whose covariance is unknown, spatially varying, and slowly time-varying. A mismatched disturbance covariance produces a persistent penalty in closed-loop stability certificates, while online estimation can reduce this penalty as data are collected. We propose a cell-wise recursive covariance estimator with spatial diffusion and prove a finite-horizon error bound that separates stochastic-approximation error, spatial-smoothing bias, and temporal-drift effects. The diffusion kernel is chosen to be reversible with respect to the stationary visitation measure, making the diffusion operator dissipative in the weighted Lyapunov analysis. We then substitute the resulting covariance estimate into the MPPI sampling distribution and derive an adaptive stability certificate with an explicit learning penalty. The main result is a payoff theorem: after a computable crossover time, the adaptive controller achieves a strictly tighter certified stability bound than any fixed covariance choice whose mismatch exceeds the residual smoothing and drift allowance. Numerical experiments illustrate the estimator convergence and the resulting stability-tightening effect.
Science Journals
arXiv:2507.18313v2 Announce Type: replace
Abstract: Malware evolves rapidly, forcing machine learning-based detectors to be continuously updated. With antivirus vendors processing hundreds of thousands of new samples daily, datasets can grow to billions of examples, making full retraining impractical. Continual learning (CL) has emerged as a scalable alternative, enabling incremental updates without full data access while mitigating catastrophic forgetting. In this work, we analyze a critical yet overlooked issue in this context: security regression. Unlike forgetting, which manifests as a drop in average performance on previously seen data, security regression captures harmful sample-level prediction changes, e.g., malware samples that were correctly detected before an update but evade detection afterward. This poses serious risks in security-critical applications, as the silent reintroduction of previously detected threats may undermine users' trust in the update process, leading them to perceive a regression in security even if the average model performance has actually improved. We first formalize and quantify security regression in CL-based malware detectors, revealing that up to 3-6% of malware experience it after model updates. We then address this issue by introducing a regression-aware framework to the CL setting. Specifically, we instantiate it via Positive Congruent Training (PCT), showing seamless integration with any prior CL strategy. Experiments on the ELSA, Tesseract, and AZ-Class datasets show that our method effectively halves regression across different CL scenarios while maintaining strong detection performance over time.
arXiv:2607.08999v1 Announce Type: new
Abstract: In large-scale computing systems, jobs often demand heterogeneous server allocations: large jobs that occupy a substantial fraction of the servers are of high importance and are thus latency-sensitive, while small jobs fill in the remaining capacity to maintain throughput. To model this dynamic, we introduce the half-batch multiserver-job (MSJ) framework, a queueing model in which large jobs arrive according to a Poisson process and require all servers simultaneously, while small jobs, each needing only one server, are always available.
We prove that, in the half-batch MSJ model, the Pareto frontier for large-job mean response time and small-job throughput admits a simple and exact characterization. It is generated by a family of convoy policies, under which the system serves small jobs until $k$ large jobs have arrived and then switches to serving large jobs, together with convex combinations of neighboring convoy policies. Our result is fully general and non-asymptotic, holding for every stable arrival rate $\lambda$, every number of servers $n$, and every large-job size distribution $S$.
arXiv:2607.09007v1 Announce Type: new
Abstract: In quantum programs, Bugs4Q is a widely used benchmark containing real quantum defects. However, its evaluation assumes that benchmark labels remain valid and that generated fixes execute in the target environment. We evaluate two Bugs4Q versions containing 67 unique real Qiskit defects, adding executable tests where missing, and re-run all entries across six pinned Qiskit releases (0.25.0, 0.45.0, 1.0.0, 1.1.1, 2.0.0, and 2.3.1). We find that quantum benchmarks can suffer from silent label inversion: entries become invalid without errors when reference fixes stop executing or buggy programs no longer reproduce failures. Thus, correctness depends on the (benchmark, version) pair rather than the benchmark alone. We evaluate four LLMs (GPT-4o-mini, GPT-5o-mini, GPT-5.4, and GPT-5.4-mini), generating up to 10 repair candidates per defect and testing them across all versions. GPT-5.4 achieves the highest pass@10 (48.8%), followed by GPT-5.4-mini (47.3%), GPT-5o-mini (30.3%), and GPT-4o-mini (22.6%). All models perform best on Qiskit 0.45.0 and decline after the Qiskit 1.0 transition. Many failures arise from deprecated or incompatible APIs rather than incorrect repairs, and 64\% of successful repairs occur on entries invalid under the target version. We release a re-validated, version-pinned Bugs4Q benchmark and show that benchmark validation must precede repair evaluation.
arXiv:2510.17170v3 Announce Type: replace-cross
Abstract: In this work, we solve a discrete optimal transport problem in a nonuniform environment. To solve the optimal transport problem, we build the cost matrix and then use classical solvers for discrete optimal transport. The challenge is to form the cost matrix, which requires finding the optimal path between two points, and for this task we formulate and solve the associated Euler-Lagrange equations. A main contribution of ours is to provide verifiable sufficient conditions of optimality of the solution of the Euler-Lagrange equation and to propose new algorithms to to check optimality a-posteriori, thus validating the (exact) computation of the cost matrix. We illustrate our results and performance of the algorithms on several numerical examples in 2 and 3 dimensions.
arXiv:2607.06327v2 Announce Type: replace
Abstract: Uncertainty estimation (UE) enables LLM-powered systems to recognize when to abstain, yet existing research has predominantly focused on English. We present the first large-scale evaluation of UE methods across 22 languages, spanning high-, mid-, and low-resource settings. Using two human-curated Q\&A datasets, we compare open and closed box UE methods (nine in total) across different model sizes and architectures while eliciting long-form reasoning, avoiding LLM-as-a-judge and embedding-based scoring, which can introduce evaluation noise. We report three main actionable findings. First, we find that prompting models to reason in English while keeping questions in low-resource languages substantially improves UE performance, suggesting that comprehension of low-resource languages is largely intact, and that the reliability bottleneck lies in generation rather than understanding. Second, prompting models to reason in English closes the UE performance gap between low and high-resource languages, demonstrating that generation language matters more than the question language. Third, the choice of UE method should depend on model scale: at smaller scales, open-box probability-based methods outperform alternatives; at larger scales, closed-box self-verbalized uncertainty becomes superior. Finally, we provide an analysis of threshold selection for selective prediction, offering guidance on calibrating abstention in multilingual settings.
arXiv:2607.08795v1 Announce Type: new
Abstract: Redundancy in non-standard numeration systems is often associated with robustness, but its practical value in finite digital arithmetic depends on how representation, storage, and repair are defined. We study intrinsic redundancy in finite beta-expansion systems using a bounded-window model that separates semantic non-uniqueness from canonical codebook admissibility. The model distinguishes arithmetic canonicalization from corruption repair and evaluates structural detectability, value-preserving repair of the observed state, and survival of the original value. For the golden-ratio system and related multinacci bases, we prove that a genuine single-digit corruption in a canonically injective finite codebook cannot be semantically recovered by exact repair without external information. Semantic survival under exact structural repair is possible only for localized multi-digit perturbations corresponding to algebraic rewrite identities, such as the equivalence of 100 and 011. Experiments comparing standard binary, signed-digit non-adjacent form, and multinacci systems quantify trade-offs among codebook sparsity, fault visibility, canonicalization cost, residual error, and boundary loss. The results show that intrinsic beta-redundancy is a constrained-language resource for structural digital integrity rather than a substitute for classical error-control redundancy.
arXiv:2607.08804v1 Announce Type: new
Abstract: As 3D software proliferates, software artifacts now extend beyond code and 2D user interfaces to include 3D assets. Among these assets, collision meshes are critical as they define the geometry used by physics engines for collision detection and physical interaction. Although existing tools can automatically generate collision meshes from visual meshes, they often fail to capture the intended interaction behavior. As a result, developers need to manually edit many heterogeneous collision meshes, a process that is time-consuming and challenging to scale.
To address this problem, we present a neuro-symbolic program synthesis approach for batch-editing collision meshes. We formulate the task as a programming-by-example problem: given a family of collision meshes with the same editing intent and a small number of user demonstrations, our approach synthesizes a reusable program that captures the editing intent and applies it to non-demonstration meshes. We implement this in a tool named MeshForge, and evaluate it across 24 tasks on 600 collision meshes. MeshForge successfully synthesizes 23/24 tasks, requiring 2.2 demonstrations and 3.5 seconds of synthesis time on average.
arXiv:2512.01317v3 Announce Type: replace-cross
Abstract: Measurement-induced entanglement (MIE) captures how local measurements generate long-range quantum correlations and drive dynamical phase transitions in many-body systems. Yet estimating MIE experimentally remains challenging: direct evaluation requires extensive post-selection over measurement outcomes, raising the question of whether MIE is accessible with only polynomial resources. We address this challenge by reframing MIE detection as a data-driven learning problem that assumes no prior knowledge of state preparation. Using measurement records alone, we train a neural network in a self-supervised manner to predict the uncertainty metric for MIE--the gap between upper and lower bounds of the average post-measurement bipartite entanglement. Applied to random circuits with one-dimensional all-to-all connectivity, our method reveals a learnability transition with increasing circuit depth: below a threshold the MIE can be effectively learned with resources that grow only polynomially with system size, whereas above it the required resources grow exponentially. This computational phase transition coincides with the breakdown of efficient classical simulation of the underlying quantum state. We further observe signatures of this transition on current noisy quantum devices. These results highlight the power of data-driven approaches for learning MIE and delineate the practical limits of its classical learnability.
arXiv:2606.30073v2 Announce Type: replace-cross
Abstract: In the control of dissipative quantum systems, the slow relaxation modes usually set the ultimate manipulation timescale. Here we show that this apparent bottleneck can be bypassed: dissipation itself becomes a control resource when fast relaxation channels are deliberately exploited. We demonstrate this mechanism for a qubit subject to non-unital and anisotropic Bloch relaxation. A short coherent pulse first reorients the Bloch vector onto a fast dissipative eigendirection; the subsequent free relaxation then carries the state close to the target, with at most one final corrective pulse. The resulting bang-drift-bang strategy is cooperative: coherent control selects the dissipative channel, while the bath performs most of the transfer. For axial targets, we obtain a closed-form speedup over passive relaxation by a factor of order $\kappa=T_1/T_2\gg1$. For out-of-equilibrium non-axial targets, an additional off-axis interception mechanism provides a further geometric amplification, allowing the hitting-time speedup, still normalized to the axial passive-reset time, to exceed the axial $\kappa\xi$ benchmark by an extra factor of four to five. The mechanism therefore directly connects to standard Bloch-vector qubit platforms, including magnetic-resonance spins, nitrogen-vacancy centers, and superconducting circuits, with potential relevance for quantum-control and fast-reset protocols.
arXiv:2607.09008v1 Announce Type: new
Abstract: Safety-critical perception systems must reliably detect rare object classes within small label spaces, a setting that long-tailed detection methods, designed for hundreds of classes with dense annotation, fundamentally do not address. Open-vocabulary detectors offer a promising alternative, as they use natural language queries at inference time, making prompt quality a first-class lever for detection performance. We exploit this property to address class imbalance: rather than retraining models or collecting additional annotations, we ask whether iteratively refining the language prompts, fed to frozen detectors, can improve minority class detection. We introduce C-GAP Caption-Guided Augmentation and Prompting), a detector-agnostic, annotation-free framework that operates in two phases. First, we establish a composite caption baseline combining per-image scene descriptions with class-quantity context, which we show outperforms scene-description only or class-quantity-only prompts across multiple open-vocabulary architectures and benchmarks. Second, an LLM iteratively refines each image's caption individually, with trials triaged into accept, tentative, or regenerate buckets based on minority-class AP@0.5 against a dynamic threshold derived from the composite baseline. Refinement terminates early once sufficient AP@0.5 gain is achieved. No detector weights are updated at any stage. Our experiments shows that C-GAP improves minority-class average precision up to 53% over the baselines. On COCO, C-GAP improves minority-class AP@0.5 by ~81% relative over the composite baseline (17.69 -> 32.09). Experiments confirm that composite captions provide the critical foundation for effective refinement: using scene-description-only or class-quantity-only prompts as the refinement starting point yields diminishing returns, supporting both stages of C-GAP as necessary contributions.
arXiv:2607.09042v1 Announce Type: new
Abstract: Reinforcement learning (RL) is increasingly used to post-train vision-language-action (VLA) models, but every update consumes robot rollouts that are slow and costly to collect, making sample efficiency a central concern. Manipulation tasks typically provide only sparse rewards, so a weak policy fails almost every rollout early in training and has little to learn from, even when those failures execute coherent behavior. Such a failure, however, is a success at a different task. We present Learning from Hindsight (LfH), which brings hindsight relabeling to RL post-training of VLAs by scoring failed rollouts against the tasks they actually achieved. A single vision-language model relabels both the instruction and the reward, proposing a hindsight instruction for a group of failed rollouts and scoring how well each satisfies it, and the policy trains on the relabeled and original rollouts jointly. Because VLAs generalize across language, relabeling in language lets the policy learn more from the same trajectories. On out-of-distribution LIBERO-PRO tasks, where standard RL improves only slowly, LfH achieves $5\times$ improvement in sample efficiency, and outperforms a dense progress-reward baseline. The gains hold across VLA backbones and on a physical Franka robot.
arXiv:2607.09136v1 Announce Type: new
Abstract: Accurate dynamics modeling of Brushless DC (BLDC) motors is fundamental to high-performance robotic joint control. This paper presents a Physics-Informed Neural Network (PINN) with a deep residual (ResNet) backbone that learns a continuous-time surrogate of the full six-state BLDC motor dynamics. Given simulation time, applied three-phase voltages, and excitation parameters as inputs, the network directly predicts all motor state variables -- rotor angle, angular velocity, three-phase currents, and winding temperature -- while simultaneously satisfying the governing electromechanical and thermal ODEs through a composite physics-data loss. A curriculum scheduling strategy gradually activates the physics penalty to prevent premature convergence. Training runs are completed in under two minutes on a standard CPU. Crucially, once trained, PINN inference achieves latencies of 0.1--22, mu s per query, up to 118x faster than conventional ODE solvers, making it suitable for real-time observer and control applications.
arXiv:2503.20659v3 Announce Type: replace-cross
Abstract: Two-dimensional PtSe2 exhibits outstanding intrinsic properties such as high carrier mobility, tunable bandgap, broadband absorption and air stability, making it ideal for (opto)electronic applications. In particular, thick PtSe2 is semimetallic and well suited for ultrafast optoelectronics in the infrared domain. However, achieving PtSe2 films of high crystalline quality with controlled properties on low-cost and insulating substrates remains challenging. Here, highly crystalline semimetallic PtSe2 films are grown by molecular beam epitaxy on sapphire substrates. It is shown how an optimized post-growth annealing remarkably improves the out-of-plane crystallinity and leads to record sheet conductances, up to 1.6 mS. In-depth structural analyses reveal the strong influence of the domain arrangement within the films on their electrical properties. Films that are mainly composed of vertically single crystalline domains exhibit high sheet conductance (1.1 - 1.6 mS), whereas films that contain superimposed twisted domains present low sheet conductance (0.5 - 0.6 mS). Moreover, it is demonstrated that the A1g Raman peak width, in addition to the commonly used Eg peak width, are both effective metrics for evaluating the quality of PtSe2: films with narrower Eg and A1g peaks exhibit higher in-plane and out-of-plane crystalline quality, respectively, as well as higher sheet conductance. Finally, coplanar waveguides integrating a semimetallic PtSe2 channel are fabricated on a 2-inch sapphire substrate to demonstrate optoelectronic devices operating at the 1.55 {\mu}m telecom wavelength. This includes photodetectors with a record 60 GHz bandwidth and the first PtSe2-based optoelectronic mixer with a bandwidth above 30 GHz.
arXiv:2505.22514v2 Announce Type: replace-cross
Abstract: In a recent Letter [H. Munoz-Bauza and D. Lidar, Phys. Rev. Lett. 134, 160601 (2025)], quantum annealing was reported to exhibit a scaling advantage in approximately solving quadratic unconstrained binary optimization (QUBO) problems. Here, we revisit these findings by employing the simulated bifurcation machine (SBM), a nonlinear dynamical system that exploits chaotic behavior rather than thermal fluctuations. Our approach originates from quantum dynamics and shares key operational features with quantum annealing: (i) nearly parallel evolution and (ii) a well-defined relation between the energy gap, run-time, and solution quality. We obtain comparable or superior scaling, closing the reported quantum-classical gap. We further show that the small instances studied previously are insufficient to infer asymptotic behavior. Extending the analysis to larger problems reveals robust classical performance, indicating that current quantum annealers are unlikely to exhibit a clear scaling advantage over SBM-like solvers on quantum-annealing-correction-type QUBO problems under the run-time accounting studied here. Finally, we identify sparse problem classes where future quantum devices could achieve a genuine scaling advantage, once hardware overheads are mitigated.
arXiv:2509.18834v3 Announce Type: replace-cross
Abstract: Microwave-optical transducers and quantum memories are essential for quantum repeaters enabling a quantum internet. Despite advances in both technologies, integrating these functionalities remains challenging. Here, we theoretically propose and experimentally demonstrate an on-demand microwave-optical quantum transducer based on a Rydberg ensemble. Using cascaded electromagnetically induced transparency, we store microwave photons in a highly excited collective state and convert them into optical photons during retrieval. Leveraging an optical depth of millions for microwave photons and minimal single-photon-level dephasing, our transducer achieves around 90\% area-normalized storage efficiency, 2.3 MHz bandwidth, and noise-equivalent temperature of 26 K under cavity-free conditions. Furthermore, our system is cryogenically compatible and extendable for high single-photon conversion efficiency without requiring optical cavity coupling. These findings advance practical on-demand quantum interfaces with broad applications across atomic and solid-state platforms.
arXiv:2510.04662v2 Announce Type: replace-cross
Abstract: Soliton gases are large ensembles of random solitons with distinct characteristics arising from integrable system dynamics. They have been widely studied in theory and experiments, and were observed in natural lagoons. However, it remains an open question whether they occur naturally in the open ocean. Nonlinear ocean states containing solitons have been observed in the literature, but the dominance of solitons over other wave components required for a soliton gas has not been demonstrated. Our study provides the first field evidence of soliton gas sea states in the deep ocean, measured in Taiwan waters. The soliton energy ratio derived from the nonlinear Fourier transform (NFT) is used as a key parameter to quantify how close sea states are to soliton gases. We identify eleven measurements with extremely high soliton energy ratios. They are characterized by short-period waves with relatively small wave heights, accompanied by extreme steepness and Benjamin Feir Index (BFI) values. These states are exceptionally rare, representing only 0.054\% of our dataset. Since directional interference can artificially increase the estimated soliton energy ratio obtained from measured time series, we further apply a probabilistic directional filtering method to remove the directional interference. Three wave records from the Eluanbi station are found to retain high soliton energy ratios after the directional interference has been removed, confirming that they are indeed soliton gases.
arXiv:2509.20005v4 Announce Type: replace-cross
Abstract: We study the activity of 'living' droplets, which confine tens of swimming meso-organisms in 3D using a superhydrophobic substrate. With few swimmers, the droplet oscillates at its inherent resonant frequency. We observe deviations from this classical regime as the level of confinement or crowding increases and develop scaling law models to successfully describe our results. We report a difference in swimming kinematics in crowded 3D environments compared to quasi-2D. Our work reveals mechanisms for bio-inspired droplet actuation with implications for mesoscale robotics, fluidics, and sensing.
arXiv:2607.09176v1 Announce Type: new
Abstract: Provenance-based attack investigation enables viable automation by standardizing data and query logic; however, it is critically hindered in practice by dependency explosions and fragmented causal chains in the wild. Towards designing a robust and automated investigation tool, we collaborated with the SOC of a major Internet corporation serving billions of users. By engaging in real-world incident response, we are able to evaluate and refine their existing LLM-based investigation workflows, which processes tens of thousands of raw alerts daily, leaving thousands for manual triage, to find out the root causes of investigation failures and major challenges in their existing tools. Motivated by these findings, we propose SherAgent, an LLM-empowered automated investigation system. Operating on an iterative ``query-filter'' backtracking paradigm over provenance graphs, SherAgent leverages the semantic reasoning capabilities of LLMs to process unstructured data, such as investigation context and threat intelligence. To overcome fragmented causal chains caused by missing events, the system dynamically calibrates query conditions to broaden the search scope. Concurrently, it performs precision result filtering and strategic nodes selection for subsequent exploration, thereby mitigating dependency explosions. Extensive evaluations in the wild demonstrate that SherAgent improves the end-to-end investigation success rate by 31.1% and 63.7% compared to both legacy enterprise baselines and SOTA approaches, respectively. Furthermore, it operates with remarkable efficiency, incurring under $0.10 in API costs and requiring less than 4 minutes per investigation. Finally, our user study confirms that SherAgent provides accurate and clear insights, significantly reducing the analytical overhead for security experts.
arXiv:2510.21742v3 Announce Type: replace-cross
Abstract: The statistics of correlations are central quantities characterizing the collective dynamics of recurrent neural networks. We derive exact expressions for the statistics of correlations of nonlinear recurrent networks in the limit of a large number N of neurons, including systematic 1/N corrections, in the regime of Gaussian quenched disorder. Our approach uses a path-integral representation of the network stochastic dynamics, which reduces the description to a few collective variables and enables efficient computation. This generalizes previous results on linear networks to include a wide family of nonlinear activation functions, which enter as interaction terms in the path integral. These interactions can resolve the instability of the linear theory and yield a strictly positive participation dimension. We present explicit results for power-law activations, revealing scaling behavior controlled by the network coupling. In addition, we introduce a class of activation functions based on Pade approximants and provide analytic predictions for their correlation statistics. Numerical simulations confirm our theoretical results with excellent agreement. We also compare with previous works that have studied the complementary case with annealed disorder, and based on this we propose a new self-consistent equation for the more general case of colored noise.
arXiv:2507.12941v4 Announce Type: replace
Abstract: Partial differential equations (PDEs) with near singular solutions pose significant challenges for traditional numerical methods, particularly in complex geometries where mesh generation and adaptive refinement become computationally expensive. Although deep-learning-based approaches, such as Physics-Informed Neural Networks (PINNs) and the Random Feature Method (RFM), offer mesh-free alternatives, they often lack adaptive resolution in critical regions, limiting their accuracy for solutions with steep gradients or singularities. In this work, we propose the Adaptive Feature Capture Method (AFCM), a novel machine learning framework that adaptively redistributes neurons and collocation points in high-gradient regions to enhance local expressive power. Inspired by adaptive moving mesh techniques, AFCM uses the gradient norm of an approximate solution as a monitor function to guide the reinitialization of feature function parameters. This ensures that partition hyperplanes and collocation points cluster where they are most needed, achieving higher resolution without increasing computational overhead. The AFCM extends the capabilities of RFM to handle PDEs with near-singular solutions while preserving its mesh-free efficiency. Numerical experiments demonstrate the method's effectiveness in accurately resolving near-singular problems with a performance that is better than that of the traditional finite element method in terms of accuracy and efficiency. AFCM offers a robust and scalable approach to solving challenging PDEs in scientific and engineering applications.
arXiv:2607.09188v1 Announce Type: new
Abstract: Agricultural machines that carry mobile tools or implements have an internal dynamic that makes them more prone to rollover risk than machines with fixed configurations. Previous research has led to the development of algorithms capable of predicting rollover risk in real time by estimating relevant metrics, primarily the LTR (Load Transfer Ratio), with a focus on vehicles with static configurations. This paper presents the adaptation of these algorithms for machines with mobile tools and describes the field tests, conducted on a self-propelled sprayer, to evaluate their effectiveness in predicting rollover. The tests also aimed to acquire the necessary data to build an accurate numerical model of the vehicle and produce high-quality simulated tests. This paper describes the chosen instrumentation, the test procedure, and the experimental design of these tests. These field tests include static tests and dynamic tests. Static tests determine vehicle parameters, while dynamic tests measure variables related to the machine displacement and wheel/ground forces in real time. The initial results demonstrate the algorithms' ability to capture variations in LTR, which express lateral and longitudinal rollover risk. The data collected through these tests will enable the vehicle displacements to be replayed by simulation, producing simulated rollover accidents and evaluating the effectiveness of the developed algorithms in accident prevention.
arXiv:2607.09190v1 Announce Type: new
Abstract: Tactile feedback is fundamental to Hand-Object Interaction (HOI), governing contact formation, force regulation, and stable manipulation, making it essential for achieving true human-like dexterous manipulation. Yet, current human-to-robot dexterous transfer pipelines primarily rely on kinematic trajectories, resulting in motion imitation without physically grounded interaction. To address this, we introduce TactiDex, a real-world tactile-guided benchmark specifically designed to move dexterous manipulation beyond kinematic mimicry toward contact-level human-likeness. TactiDex provides a comprehensive dataset that elegantly aligns whole-hand tactile signals with multi-granularity kinematic and object states, coupled with standardized evaluation metrics. Building upon this data paradigm, we propose a tactile-driven transfer framework that effectively translates human demonstrations into physically plausible robotic execution. We introduce TactiSkill, a framework built upon a novel tri-component tactile reward that innovatively uses tactile signals as structured supervision. This reward unifies guidance, human-like alignment, and contact constraints into a single objective. Through comprehensive experiments on both single and bimanual tasks, we demonstrate that TactiSkill achieves superior performance in manipulation success and physical realism. This work lays a crucial foundation for advancing tactile-aware dexterous manipulation. Our project page at https://tactidex.github.io/.
arXiv:2602.21479v3 Announce Type: replace-cross
Abstract: Across many risk-sensitive areas, it is critical to continuously audit machine learning systems as we receive more data to quickly determine if they are performing as designed. This auditing task can be modeled as a sequential hypothesis testing problem with $k$ data streams and a global null hypothesis that asserts the system operates as intended across all $k$ streams. Under the alternative, the standard global sequential test, which uses a Bonferroni correction, has an expected stopping time of $O\left(\ln \frac{k}{\alpha}\right)$ for large $k$ and significance level $\alpha$. In this work, we demonstrate that efficient sequential tests, relying on merging martingales via averaging and products rules, provide improved stopping times, and thus more powerful tests against the null. Using these results, we show that a balanced test can match the Bonferroni rate of $O\left(\ln \frac{k}{\alpha}\right)$ in the sparse regime (just a few non-null streams) while achieving $O\left(\frac{1}{k}\ln \frac{1}{\alpha}\right)$ under dense alternatives (many non-null steams). We validate our theory through experiments on both synthetic and real-world data.
arXiv:2607.09210v1 Announce Type: new
Abstract: Perceived social support can buffer stress, but how it is associated across online social networks at different graph distances remains unclear. Here we show that inferred perceived online social support in a large avatar communication application decays with network distance in a form better described, over the observed range, by a power-decay model than by a single exponential. We linked two-wave survey data from Pigg Party with behavioral logs, trained a random forest to infer perceived support for active users, and regressed Wave 2 scores on Wave 1 scores for users at hop distance $k$, adjusting for baseline support and covariates. Adjusted associations persisted across hops, consistent with power-law-like decay over the observed range. Individual-based simulations indicated that heterogeneous source-specific exponential decay rates can generate heavy-tailed aggregate decay. These results suggest that network position heterogeneity should be considered when characterizing distance-dependent associations among psychosocial states in online communities.