arXiv:2504.17070v3 Announce Type: replace Abstract: Robots need task planning methods to achieve goals that require more than one action. Recently, large pretrained models have demonstrated impressive performance in task planning. For instance, large language models (LLMs) can generate task plans using action and goal descriptions. Despite the rapid progress of large models in robot intelligence, their security implications remain only partially understood, leaving important gaps in the exploration of potential vulnerabilities in LLM-driven robotic planning systems. To investigate such risks, in this paper, we develop MuTRAP, the first multi-trigger trojan attack specifically designed and targeted for LLM-assisted robot task planners. MuTRAP follows the standard practice of LLM usage in robotics where the backbone LLM is typically frozen and hosted in a central server limiting attacker's reach. In contrast, MuTRAP injects backdoor using a small set of task-specific parameters. In addition, we develop a trigger optimization method for selecting multiple-trigger words that are most effective for different robot applications. For instance, one can use unique trigger word "herical" to activate a specific malicious behavior, e.g., cutting hand on a kitchen robot. Through MuTRAP that demonstrates the vulnerability of current LLM-based planners, our goal is to promote the development of secured robot intelligence. Details and demos are provided in: https://mutrap.github.io/MuTRAP/
Science Journals
arXiv:2509.03647v2 Announce Type: replace Abstract: Large language models (LLMs) increasingly serve as automated evaluators, yet they suffer from "self-preference bias": a tendency to favor their own outputs over those of other models. This bias undermines fairness and reliability in evaluation pipelines, particularly for tasks like preference tuning and model routing. We investigate whether lightweight steering vectors can mitigate this problem at inference time without retraining. We introduce a curated dataset that distinguishes self-preference bias into justified examples of self-preference and unjustified examples of self-preference, and we construct steering vectors using two methods: Contrastive Activation Addition (CAA) and an optimization-based approach. Our results show that steering vectors can reduce unjustified self-preference bias by up to 97\%, substantially outperforming prompting and direct preference optimization baselines. Yet steering vectors are unstable on legitimate self-preference and unbiased agreement, implying self-preference spans multiple or nonlinear directions. This underscores both their promise and limits as safeguards for LLM-as-judges and motivates more robust interventions.
arXiv:2606.24429v1 Announce Type: new Abstract: Generative AI coding agents are entering the open-source supply chain, yet their diverse and often invisible traces leave their prevalence poorly understood. We introduce a multi-layered detection framework that integrates configuration-file scanning, commit-message analysis, author-identity matching, and bot-signature lookup across World of Code (180M+ Git repositories), classifying agent traces into four behavioral types. No single method captures more than a fraction of activity: multi-method detection identifies 850,157 Claude Code commits in one snapshot, of which bot-account lookup_the signal most adoption studies rely on_recovers only 28,154 (3.3%), a 30x relative-recall gap, so single-signal prevalence estimates are biased low by at least this factor. Every detection pattern is hand-validated (495 labels) with per-cell precision and Wilson confidence intervals. Across snapshots from December 2024 to April 2026, commit-attributed agents generate over 320,000 commits per month; Claude Code leads (886,122 commits across 17,295 projects) and dominates silent, configuration-file-only adoption (21,078 projects). Compared against an independent pull-request census (AIDev), the two channels capture nearly disjoint agent populations_a PR census misses 79% of commit-detected Claude Code adopters and essentially all Codex adopters_and different kinds of work: PR-deployed cloud agents (Codex, Cursor) surface as feature work, while commit-deployed in-editor agents (Claude Code, OpenHands, Aider) surface as maintenance. The observed work profile follows deployment and detection mode rather than the tool itself, so no single channel is representative.
arXiv:2504.17768v3 Announce Type: replace Abstract: Sparse attention offers a promising strategy to extend long-context capabilities in Transformer LLMs, yet its efficiency-accuracy trade-offs remain unclear due to the lack of comprehensive evaluation. We address this gap with the largest-scale empirical analysis to date of training-free sparse attention, evaluating six methods across multiple model families and sizes, sequences up to 128K tokens, and sparsity levels up to 0.95 (i.e., $1/20$ attention budget) on nine diverse tasks. We first organise the rapidly evolving landscape of sparse attention methods into a taxonomy along four design axes. Our analysis then yields actionable insights: 1) sparse attention is effective: larger sparse models outperform smaller dense ones at equivalent cost, improving the Pareto frontier; 2) for the training-free methods we study, fine-grained per-query importance estimation during prefilling remains impractical-due to both the cost of estimation and the lack of sparse kernels that translate fine-grained sparsity into wall-clock gains-forcing a task-dependent choice between global-to-token and block-to-block selection. Instead, during decoding, token-to-page selection becomes feasible, enabling better generalisation and higher sparsity tolerance; 3) longer sequences tolerate higher sparsity, suggesting that fixed-budget methods in production are suboptimal. Together, these findings provide practical guidance for deploying sparse attention and methodological recommendations for future evaluations. Our code is available at https://github.com/PiotrNawrot/sparse-frontier.
arXiv:2505.10457v3 Announce Type: replace Abstract: Incremental learning is a machine learning paradigm where a model learns from a sequential stream of tasks. This setting poses a key challenge: balancing plasticity (learning new tasks) and stability (preserving past knowledge). Neural Architecture Search (NAS), a branch of AutoML, automates the design of the architecture of Deep Neural Networks and has shown success in static settings. However, existing NAS-based approaches to incremental learning often rely on expanding the model at every task, making them impractical in resource-constrained environments. In this work, we introduce SEAL, a NAS-based framework tailored for data-incremental learning, a scenario where disjoint data samples arrive sequentially and are not stored for future access. SEAL adapts the model structure dynamically by expanding it only when necessary, based on a capacity estimation metric. Stability is preserved through cross-distillation training after each expansion step. The NAS component jointly searches for both the architecture and the optimal expansion policy. Experiments across multiple benchmarks demonstrate that SEAL effectively reduces forgetting and enhances accuracy while allocating additional capacity only when required. These results highlight the promise of combining NAS and selective expansion for efficient, adaptive learning in incremental scenarios.
arXiv:2606.07469v2 Announce Type: replace-cross Abstract: This paper sets out the most general computational and econometric implications of the rigorous stochastic equilibrium theory from SELCKE (Staines (2024a)) arXiv:2312.16214. The analytical backbone is the discovery that the system converges geometrically to long-run equilibrium, at a rate given by the greater of the eigenvalue or inverse eigenvalue (from outside) closest to the unit circle and the maximum shock persistence. High-order shocks converge faster. I develop a simulation procedure to test, with asymptotic power, whether stochastic equilibrium exists for a particular model. The fundamental approximation result asserts that, whatever the order of expansion or loss function, the stochastic steady state delivers the most accurate perturbation solution. I also show that super-consistent parameter estimators $O(1/T)$ arise whenever second-order terms vanish. Besides Calvo, I study stochastic equilibrium in two alternative pricing models. Dynamics simplify considerably. I bound the time the impulse response peaks, by the maximum lag in the errors. This lends empirical support to Taylor contracts, although there are issues surrounding unit roots and the strong cost-channel. For menu costs, I demonstrate that the initial price distribution decays away super-exponentially, producing a system equivalent to Calvo with an endogenous reset probability. The impact of idiosyncratic disturbances appears as an additional wedge between actual and efficient output. Blow-up of the objective function at the boundary is proven, with the help of new distributional arguments, so the model meets existing eigenvalue existence conditions for the recursive equilibrium. Along the way, new light is shone on existing theoretical models and statistical procedures.
arXiv:2505.10931v4 Announce Type: replace Abstract: Single-source remote sensing object detection using optical or SAR images struggles in complex environments. Optical images offer rich textural details but are often affected by low-light, cloud-obscured, or low-resolution conditions, reducing the detection performance. SAR images are robust to weather, but suffer from speckle noise and limited semantic expressiveness. Optical and SAR images provide complementary advantages, and fusing them can significantly improve the detection accuracy. However, progress in this field is hindered by the lack of large-scale, standardized datasets. To address these challenges, we propose a new comprehensive dataset for optical-SAR fusion object detection, named Multi-resolution, Multi-polarization, Multi-scene, Multi-source SAR dataset (M4-SAR). It contains 112,174 instance-level aligned image pairs and nearly one million labeled instances with arbitrary orientations, spanning six key categories. To enable standardized evaluation, we develop a unified benchmarking toolkit that integrates six state-of-the-art multi-source fusion methods. Additionally, we propose E2E-OSDet, a novel end-to-end multi-source fusion detection framework that mitigates cross-domain discrepancies and establishes a robust baseline for future studies. Extensive experiments on M4-SAR demonstrate that fusing optical and SAR data can improve mAP by 5.7\% over single-source inputs, with particularly significant gains in complex environments. The dataset and code are publicly available at https://github.com/wchao0601/M4-SAR.
arXiv:2505.13087v2 Announce Type: replace Abstract: We propose a novel benchmarking methodology for graph neural networks (GNNs) based on the graph alignment problem, a combinatorial optimization task that generalizes graph isomorphism by aligning two unlabeled graphs to maximize overlapping edges. We frame this problem as a self-supervised learning task and present several methods to generate graph alignment datasets using synthetic random graphs and real-world graph datasets from multiple domains. For a given graph dataset, we generate a family of graph alignment datasets with increasing difficulty, allowing us to rank the performance of various architectures. Our experiments prove that there is an optimal task difficulty for having a statistically relevant ranking of different models and that, even on a structure-only task, anisotropic models perform better compared to isotropic ones. To further prove that our synthetic task capture meaningful information, we show its effectiveness for self-supervised GNN pre-training: the learned node embeddings can be leveraged as positional encodings by transformers for graph regression or can be used to reconstruct the full structure of the graph with $98\%$ accuracy. To support reproducibility and further research, we provide an open-source Python package to generate graph alignment datasets and benchmark new GNN architectures. The source code is available at https://github.com/adrien-lagesse/graph-alignment-benchmark.
arXiv:2606.24433v1 Announce Type: new Abstract: Medical point cloud completion is important for anatomical reconstruction and downstream clinical workflows, yet generative modeling in this setting remains insufficiently studied. We investigate completion through continuous-time generative modeling and introduce PCFM, a PTv3-backed flow matching approach for medical point cloud completion. We evaluate on SkullFix and SkullBreak, and additionally on the more recent Mandibular Defect dataset. We build strong baselines by adapting PTv3 to a deterministic encoder-decoder completion model and by instantiating diffusion completion (PCDiff) with both PVCNN and PTv3 denoisers. PCFM with PTv3 is competitive with the deterministic PTv3 baseline and achieves state-of-the-art generative performance across datasets, while requiring substantially fewer sampling steps than diffusion. At the best operating points, PTv3 also yields clear throughput gains, providing up to a 7$\times$ speed-up for PCFM compared to a PVCNN backbone. Finally, we study empirical scaling trends by varying model size and point cardinality, showing consistent gains with higher point resolution and informative trade-offs across model scales.
arXiv:2606.24074v1 Announce Type: new Abstract: Wang~\cite{Wang2026} introduced the Stochastic-Oracle Turing Machine (SOTM) framework and defined token complexity as the minimum expected cost of interacting with a stochastic oracle needed to attain a specified solution quality for a task. This paper develops an analogous notion for certifying the reliability of a stochastic oracle on a given domain. Certification token complexity is the minimum expected token cost required, with controlled error probability, to distinguish oracles that meet a target reliability level from those that fall below a lower reliability threshold. We construct an SPRT-based certification SOTM that queries the oracle, computes binary correctness scores, and stops when the accumulated log-likelihood evidence crosses a decision threshold. The SOTM halts almost surely, satisfies the desired two-sided error guarantee over the reliability regions to be certified, and yields an explicit upper bound on certification token complexity in terms of the reliability thresholds, the error bound, and the expected per-turn token cost. We then establish a matching information-theoretic lower bound: even with adaptive queries, every error-bounded certification SOTM must incur the same leading-order expected token cost as the SPRT-based construction as the prescribed error bound tends to zero. Together, these bounds characterize the leading-order certification token complexity in the small-error regime.
arXiv:2505.18542v4 Announce Type: replace Abstract: Extracting structured procedural knowledge from unstructured business documents is a critical yet unresolved bottleneck in process automation. While prior work has focused on extracting linear action flows from instructional texts, such as recipes, it has insufficiently addressed the complex logical structures, including conditional branching and parallel execution, that are pervasive in real-world regulatory and administrative documents. Furthermore, existing benchmarks are limited by simplistic schemas and shallow logical dependencies, restricting progress toward logic-aware large language models.To bridge this Logic Gap, we introduce BREX, a carefully curated benchmark comprising 409 real-world business documents and 2,855 expert-annotated rules. Unlike prior datasets centered on narrow service scenarios, BREX spans over 30 vertical domains, covering scientific, industrial, administrative, and financial regulations. We further propose ExIde, a structure-aware reasoning framework that investigates five distinct prompting strategies, ranging from implicit semantic alignment to executable grounding via pseudo-code generation. This enables explicit modeling of rule dependencies and provides an out-of-the-box framework for different business customers without finetuning their own large language models. We benchmark ExIde using 13 state-of-the-art large language models. Our extensive evaluation reveals that executable grounding serves as a superior inductive bias, significantly outperforming standard prompts in rule extraction. In addition, reasoning-optimized models demonstrate a distinct advantage in tracing long-range and non-linear rule dependencies compared to standard instruction-tuned models.
arXiv:2505.21916v3 Announce Type: replace Abstract: Embodied robots have achieved strong performance in many real-world manipulation tasks, yet agile dynamic manipulation remains challenging due to high sensitivity to motion parameters and sparse outcome-level feedback. Tasks such as shooting a basketball into a hoop require precise control of fast open-loop motions, where small trajectory variations can lead to large outcome deviations, making data-efficient adaptation difficult for existing methods that rely on large-scale interaction, reward engineering, or accurate dynamic modeling. We propose Prior Reinforce (P.R.), a simple and practical framework for goal-conditioned dynamic manipulation. The method first learns a structured motion manifold from a small set of demonstrations using a conditional diffusion model, and then adapts motions toward new goals through feedback-driven optimization in a low-dimensional condition space. By separating motion generation from outcome-driven adaptation, the framework enables efficient refinement using only a small number of real-world trials under noisy perception. Experiments on multiple real-world dynamic manipulation tasks demonstrate that P.R. reliably achieves new goals within as few as ten total trials while remaining robust to perception noise and hardware uncertainty, suggesting a practical approach for low-trial real-world robot adaptation. Project website: https://adap-robotics.github.io/.
arXiv:2505.24622v3 Announce Type: replace Abstract: Many high-stakes screening tasks require predicting rare outcomes from unstructured text, where errors are costly and decisions must be auditable. We introduce Random Rule Forest (RRF), an interpretable ensemble that uses a large language model (LLM) not as an end-to-end predictor but as a generator of simple YES/NO questions. Each question acts as a weak learner, and their responses are combined by a plain unit-weight vote into an auditable ``green-flags'' scorecard: enough independent positive signals indicate a higher chance of success. We argue this deliberate simplicity is a robust default when positives are scarce and learned weights are hard to estimate. We evaluate RRF in two low-base-rate domains. On early-stage startup screening from founder profiles, RRF produces a transparent scorecard whose precision is several times the base rate (with light expert input raising it further) and, unlike direct prompting, its operating point can be controlled directly. On an established Phase~I clinical-trial benchmark, RRF outperforms published baselines on the threshold-independent metrics PR-AUC and ROC-AUC. Together these show that LLMs can serve as auditable feature generators for high-stakes text-based decisions, combining transparency with competitive predictive performance.
arXiv:2606.16821v2 Announce Type: replace Abstract: Large language model (LLM)-based search agents synthesize open-web content into actionable recommendations on behalf of users, creating a risk that attacker-published pages are transformed into endorsed claims. We introduce SearchGEO, a controlled evaluation framework for measuring endorsement corruption in LLM-based web-search agents, combining a web-evidence manipulation pipeline, a five-mode attack taxonomy, and multiple output-level metrics. We evaluate 13 LLM backends on 308 cases each. Results show that vulnerability patterns vary across backends: overall attack success rate (ASR) ranges from 0.0% on Claude-Sonnet-4.6 to 31.4% on Gemini-3-Flash, the strongest attack mode differs by model family, and the same deployment scaffold could amplify or decrease ASR on different backends. An auxiliary agent-skill probe, where endorsement becomes an install command, exposes a sharp split among otherwise robust backends: Claude over-rejects while GPT over-trusts. These findings argue for treating recommendation reliability under adversarial search content as a first-class dimension of backend safety evaluation.
arXiv:2606.24471v1 Announce Type: new Abstract: We prove that random linear codes have nearly optimal discrepancy properties in a broad range of regimes. Our main results are two general theorems: one controlling all translates of a fixed test, and another controlling large families of Fourier-pseudorandom tests. Two motivating applications follow. First, random linear codes match unstructured random codes for list-decoding from errors above capacity. If $C\subseteq\mathbb F_q^n$ is a random linear code of rate $1-\frac1n\log_q |B_\rho|+\epsilon$, where $B_\rho$ is a radius-$\rho$ Hamming ball, then with high probability $$ |C\cap B|=(1\pm o(1))\frac{|C||B|}{q^n} $$ simultaneously for all radius-$\rho$ Hamming balls $B\subseteq\mathbb F_q^n$. This extends the classical result that such codes have covering radius at most $\rho n$ whp (Blinovsky, 1987). Second, over prime fields, random linear codes match unstructured random codes for zero-error list-recovery above capacity. For prime $q>2$ and $2\le \ell\le q-1$, a random linear code of rate $1-\log_q\ell+\epsilon$ satisfies, with high probability, $$ |C\cap S|=(1\pm o(1))\frac{|C|\ell^n}{q^n} $$ simultaneously for all rectangles $S=S_1\times\cdots\times S_n$ with $|S_i|=\ell$. As a consequence, there are abundant $n$-party linear ramp secret sharing schemes over $\mathbb F_q$ with privacy threshold about $n/(2\log q)$ and reconstruction threshold about $5n/(2\log q)$, resilient to balanced local leakage; prior existence results required thresholds above $n/2$ even in this case. The translate result, hence the list-decoding application, holds over arbitrary finite fields, even growing with $n$. The list-recovery and leakage applications hold over prime fields under moderate growth, e.g. $q\le n^{1/5-o(1)}$. The proofs use a refined second-moment analysis tracking intersection sizes as random generators are added to $C$.
arXiv:2507.09839v2 Announce Type: replace Abstract: An increasing number of NLP applications interact with large language models (LLMs) through black-box APIs, making prompt engineering critical for controlling model behavior. Recent Automatic Prompt Optimization (APO) methods iteratively refine prompts using model-generated critiques (often called textual gradients), but they predominantly optimize from failures and underutilize information contained in correct predictions, leading to instability and semantic drift. We propose TRAS (Textual Regularization with Aggregated Signals), a feedback-centric framework that is plug-and-play with existing APO search backbones. It retains the standard textual gradient signal from prior work for error correction and introduces a complementary textual regularizer derived from successful predictions to preserve beneficial prompt components. Because both signals are stochastic and can be noisy, we further introduce Monte Carlo Signal Aggregation (MCSA), which samples multiple gradients or regularizers and aggregates them into a single actionable directive, emphasizing consistent, actionable advice while filtering out outliers. Motivated by rapid model churn, we also formalize Automatic Prompt Migration (APM), the practical problem of adapting an expert prompt across model versions or API providers without losing critical instructions. Across standard APO and APM scenarios, our approach consistently outperforms strong baselines, yielding higher accuracy, faster convergence, and lower query cost, while substantially reducing the degradation observed under naive prompt migration.
arXiv:2606.23704v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated capabilities in electronic design automation (EDA) for integrated circuits. However, their applications in printed circuit board (PCB) design and analysis tasks remain underexplored. In part, this is due to a (1) a lack of text-based PCB datasets to evaluate LLMs and (2) a lack of methodologies for prompting LLMs with different types of PCB design files. To address this gap, our paper proposes PCB-QA: a manually created questionnaire dataset amounting to 480 question-answer pairs for PCBs, derived from 8 different open-source hardware projects of varying complexities. We examine multiple aspects of PCB designs and cover questions about component connections, datasheet examination, and simulation data obtainable via SPICE. Using our dataset as a benchmark, we prompt LLMs with different representations (and combinations) of PCB design files and record observations. This allows us to measure, for the first time, if LLMs can understand schematics and netlists in their "native forms" (i.e. graphical PDFs, KiCAD-format design files) or if textual formats are preferred. Including both commercial and open-weight models, we benchmark 4 state-of-the-art LLMs on our dataset, finding that Gemini 3 Flash Preview can answer questions with an accuracy of 93% using a proposed JSON-based textual format. This demonstrates that text-based PCB design formats can be evaluated by LLMs. Our open-source questionnaire is the first step towards enabling LLM integrations within the PCB design life cycle.
arXiv:2606.23709v1 Announce Type: new Abstract: Integrated sensing and communication (ISAC) in cell-free (CF) massive multi-user multiple-input multiple-output (MU-MIMO) system is a promising architecture for high-rate communications and high-accuracy multi-target sensing. However, centralized coordination among distributed access points (APs) incurs substantial fronthaul overhead and computation complexity. This paper proposes a low-complexity hybrid precoding framework for CF massive MU-MIMO ISAC systems with partially-connected architectures at the APs. By applying hybrid architecture at the APs, the proposed framework converts the original high-dimensional channel information into a low-dimensional effective channel, enabling digital precoding over the compressed channel domain and thereby substantially reducing both fronthaul overhead and baseband computational complexity. We formulate the joint hybrid precoding design as an ergodic sum-rate (ESR) maximization problem with position error bound (PEB) constraints to ensure multi-target sensing accuracy. An efficient alternating optimization (AO)-based solver is then developed, where the PEB constraint is reformulated into tractable convex constraints, while the digital-domain optimization is carried out over the reduced-dimensional effective channel and the analog precoding is refined on the constant-modulus manifold. For dynamic user topology, we further propose multi-branch (MB) rate-splitting (RS) minimum mean-square-error Tomlinson-Harashima precoding (MMSE-THP) update algorithm that combines multi-branch ordering with recursive MMSE-THP matrix updates, enabling common and private digital precodings to be refreshed without repeated full matrix recomputation. Simulation results demonstrate that the proposed scheme achieves high ESR and accurate multi-target sensing while reducing computational complexity by 87.02\% compared with conventional baselines.
arXiv:2606.23716v1 Announce Type: new Abstract: Legal AI benchmark research frequently invokes the assumption that large language models can improve access to justice, including for people who cannot access lawyers in order to understand and exercise their legal rights. We argue that current benchmarks are not equipped to support this assumption because they evaluate legal reasoning over inputs that have already been preprocessed by legal experts, which measures the upper bound of model performance. Access to justice depends on a lower bound: how models perform when inputs come from pro se litigants, whose prompts may contain noisy narratives, buried facts, omissions, folk-legal assumptions, and surface-level errors. These degradations are comparable to conditions under which LLMs are known to degrade in the general machine learning literature, including long-context sensitivity, underspecification, hallucination, and typographical perturbations. We connect evidence from pro se literature with this body of machine learning research and present a small perturbation experiment on LEXam, a legal benchmark, to illustrate the gap between these two bounds. If model development continues to focus on benchmarks that measure only the upper bound, this gap may remain hidden or even widen. We conclude by calling for legal benchmarks that directly measure robustness under pro se-like inputs so that access-to-justice claims about legal AI can become empirically testable.
arXiv:2606.23724v1 Announce Type: new Abstract: Large language models are increasingly used to answer questions over annual reports, earnings decks, and analyst notes, yet their outputs remain difficult to verify in high-stakes financial workflows. A fluent answer can blend directly grounded statements, weak synthesis, and unsupported claims across narrative text, tables, and charts. We present EvidenceLens, a visual analytics prototype that treats financial question answering as a claim-evidence alignment problem. The system decomposes an answer into atomic claims, summarizes support composition and confidence, support gaps, and coordinates claim-level inspection with source passages, table cells, and chart regions. Its core visual representation is a multimodal claim-evidence matrix that makes coverage, contradiction, and modality imbalance immediately visible. To support reproducibility, we also specify a JSON-based artifact schema, a lightweight multimodal alignment pipeline, and a deterministic review-priority ranking that maps backend signals into an auditable visual structure. Through representative report-auditing scenarios, we show how EvidenceLens helps analysts distinguish grounded claims from overconfident synthesis that conventional chat interfaces flatten.
arXiv:2512.00324v4 Announce Type: replace Abstract: Dexterous robotic hands are expected to perform complex, contact-rich object manipulation, but learning such skills remains challenging because high-dimensional hands require high-fidelity demonstrations. Imitation learning provides a practical route for acquiring dexterous manipulation skills from human demonstrations, yet collecting synchronized multimodal demonstrations with accurate hand actions and tactile observations remains a key bottleneck. We present MILE, a teleoperation-based data-collection system comprising the human-first MILE exoskeleton and the mechanically corresponding MILE-Tac robotic hand. The system integrates custom-designed and fabricated modular joint encoders and compact MILE fingertip visuotactile sensor modules. The exoskeleton is informed by human-hand anatomy and ergonomic constraints, while the robotic hand is co-designed to preserve the selected four-finger kinematic topology. This correspondence enables joint-space command transfer and reduces reliance on task-space IK-based retargeting. The system synchronously records task-specific visual observations, four fingertip visuotactile streams, robot-hand proprioception, and exoskeleton-derived action commands. We evaluate MILE through a four-task teleoperation benchmark against representative glove-based and vision-based interfaces, and through imitation-learning experiments that compare policies trained with and without fingertip tactile input. The project page is available at https://sites.google.com/view/mile-system.
arXiv:2512.08766v3 Announce Type: replace Abstract: Zermelo's navigation problem seeks the trajectory of minimal travel time between two points in a fluid flow. We address this problem for an agent -- such as a floating drone or active particle -- that is advected by a two-dimensional flow, self-propels at a fixed speed smaller than or comparable to the characteristic flow velocity, and can steer its direction. The flows considered span increasing levels of complexity, from steady solid-body rotation and time-dependent sink-vortex to the Taylor-Green flow and turbulence in the inverse energy cascade regime. Although optimal-control theory provides time-minimizing trajectories, these solutions become unstable in chaotic regimes characterized by positive finite-time Lyapunov exponents. To design robust navigation strategies, we apply reinforcement learning and compare Q-learning with a one-step actor-critic algorithm. Both methods achieve successful navigation, yielding mean travel times within 3-10% of optimal-control solutions in regular flows, while the discrepancy increases to 35-75% in time-dependent turbulent flows. Finally, we show that agents trained on coarse-grained turbulent flows generalize to the full velocity field. This robustness to incomplete flow information is essential for practical navigation in real-world oceanic and atmospheric environments.
arXiv:2512.21835v2 Announce Type: replace Abstract: Providing lossless inference services of LLMs on edge devices remains challenging, especially given the extremely tight memory budgets. The existing offloading techniques inevitably introduce numerous loading bubbles, which further inflate the end-to-end latency of the entire inference pipeline. Meanwhile, dynamically fluctuating network bandwidth and diverse user request patterns pose additional obstacles to efficient lossless inference on edge devices. To address this, we propose LOIP, a collaborative lossless LLM inference system that employs an offloading-based interleaved pipeline parallelism to better overlap model offloading with computing and communicating. Specifically, LOIP first constructs an offloading-aware cost model to characterize inference latency and memory overhead under heterogeneous device capabilities and limited bandwidth. Based on this cost model, LOIP develops a fine-grained allocation scheduler that determines latency-efficient layer partitions across devices while explicitly accounting for offloading overhead, along with a unified memory architecture (UMA)-aware loading optimization using customized CUDA operators to reduce runtime loading overhead. LOIP further designs an online memory adaptation strategy to handle the increasing KV cache pressure and dynamic bandwidth fluctuations during inference. We implement LOIP with 2500+ lines of Python and 500+ lines of C++/CUDA code, and deploy it on five heterogeneous NVIDIA Jetson edge devices for lossless collaborative inference of LLaMA3.3-70B-Instruct. Extensive experiments demonstrate that LOIP achieves 8.8$\times$$\sim$20.3$\times$ speedups over the SOTA baselines under different bandwidth conditions and request patterns without compromising model accuracy.
arXiv:2601.04820v2 Announce Type: replace Abstract: Time series decomposition into trend, seasonal, and residual components is a fundamental primitive in data mining and analytics pipelines, underpinning anomaly detection, change-point analysis, and forecasting. Most existing methods require a user-specified or estimated season length and assume stable periodic structure. In large, heterogeneous collections, where recurring patterns drift, appear intermittently, or operate at multiple nonstationary scales, period selection becomes brittle and per-series tuning does not scale. We propose LGTD (Local-Global Trend Decomposition), a season-length-free decomposition framework that requires no period specification and operates with a single fixed default configuration across datasets. LGTD represents a series as the sum of (i) a smooth global trend capturing long-term evolution, (ii) adaptive local trends inferred by an error-driven local linear segmentation procedure, and (iii) a residual component. Rather than modeling seasonality through an explicit periodic basis, LGTD treats it as an emergent property arising from the recurrence of local trend regimes, decoupling decomposition quality from any estimated season length. We prove that the local trend inference procedure terminates in a bounded number of iterations and runs in linear time in the series length, independent of any seasonal parameter, and confirm this empirically: LGTD scales linearly in runtime and memory and is the fastest method across all tested lengths, while several baselines degrade super-linearly. On synthetic benchmarks LGTD achieves balanced accuracy across fixed, transitive, and variable season-length regimes, particularly where period-based methods degrade, and on real-world data it yields interpretable components and low-structure residuals. Source code and datasets are available at https://github.com/chotanansub/LGTD.
arXiv:2601.04898v2 Announce Type: replace Abstract: Phase-field (PF) modeling is a powerful tool for simulating microstructure evolution. To accelerate the simulation of PF models governed by complex PDEs, machine learning methods such as PINNs and ConvLSTM have been introduced. However, current machine-learning-based approaches still suffer from limited flexibility, poor generalization, and short prediction horizons. To address these challenges, we present a joint framework that couples a voxel-flow network (VFN) with PF simulations in an alternating manner for long-horizon prediction of microstructure evolution with substantial computational acceleration. The VFN iteratively predicts future evolution by generating the next snapshot from the previous two snapshots. Periodic PF simulations suppress nonphysical artifacts, reduce accumulated error, and extend the reliable prediction horizon. The VFN was validated using a grain-growth example, and its accuracy outperforms that of similar prediction methods while preserving topological grain details. For an ultra-long grain-growth prediction of 82 frames from 2 input frames, the grain number decreases from 600 to 29 while the NMSE of the average grain area remains 1.64%. The framework also exhibits good generalizability across different PF models. Overall, this joint framework enables rapid, flexible, generalizable, and physically consistent microstructure forecasting from image-based data over ultra-long time scales.