Papers

Semantic communication has recently gained traction for its ability to reduce the amount of data transmitted over a communication link by transmitting a task-oriented representation instead of the raw source. Zero-shot semantic communication sends a general embedding from a vision-language model (VLM), so the same transmitter can serve new classification tasks without retraining. Most evidence for this advantage, however, comes from numerical simulation. We implement zero-shot semantic communication on a software-defined radio platform: a Raspberry Pi drives a pair of Analog Devices Active Learning Module (ADALM)-Pluto transceivers, with an image encoder at the transmitter and a text encoder at the receiver, and determines the zero-shot classification results via cosine similarity. We compare two VLMs, CLIP and MobileCLIP, across various channel conditions, i.e., different signal-to-noise ratios (SNRs). We validate that the semantic link spends 9x fewer channel uses per image than a JPEG plus 16-ary quadrature amplitude modulation baseline and still reaches 82% accuracy on CIFAR-10 at 22.3 dB, where the baseline scores 0%. On the traffic sign recognition dataset (TSRD), MobileCLIP correctly classifies 98.3% of unseen images at the same SNR. Offloading the image encoder to a neural processing unit reduces encoding to 13.4 ms per image, 49x faster than a Raspberry Pi 4 CPU, placing the transmitter within a real-time budget. Our implementation is publicly available at https://github.com/thanhlexyz/zsscsdr.

In this paper, we introduce a class of function-preserving neural network operators designed to preserve a prescribed strictly positive function. We examine the approximation behaviour of the proposed operators on the space of continuous functions and show that they are positive, linear, well-defined, and bounded. Specifically, we demonstrate uniform convergence on compact intervals and get quantitative estimates of the approximation error in terms of the modulus of continuity. To support the theoretical analysis, we present several numerical experiments focusing on the preservation of exponential and logarithmic functions. These examples show how well the proposed operators perform and compare them with existing neural network operators. Also, an application to denoising modulated signals is provided, demonstrating the practical relevance of the proposed approach.

Holes in LiDAR scans of environments with glass windows remain an unresolved problem in three-dimensional reconstruction. This study presents a pipeline for point cloud acquisition, filtering, completion, and surface reconstruction to address sparse sampling and missing window regions in scans of a high-speed train nose. FAST-LIVO2 provides the initial point cloud through multisensor odometry and mapping, and moving least squares (MLS) smooths the observations. We then introduce three-axis projection-based subdivision and interpolation with reverse hole boundary identification, referred to as three-axis reverse completion. The method interpolates missing regions from observations around each hole. Greedy projection triangulation, Poisson surface reconstruction, and a Marching Cubes-based pipeline generate meshes from the completed point cloud. Experiments on a proportionally scaled display model of a high-speed train nose show that the proposed method fills missing point cloud regions around the glass windows. Under the evaluation setting used in this study, greedy projection triangulation yields lower geometric distance errors than the other two reconstruction pipelines. The pipeline supports non-contact digital modeling of train nose geometry and provides a practical approach to reconstructing objects with glass windows.

This paper introduces Hybrid DeepSEE (HDS), a Human-in-the-Loop (HITL) neuro-symbolic framework for proactive drift anticipation in Visual SLAM (V-SLAM). While data-driven models offer predictive power, their "black-box" nature often yields physically inconsistent outputs in out-of-distribution (OOD) environments. To address this, HDS integrates neural drift risk estimation with symbolic constraint reasoning. By utilizing a Large Language Model (LLM) as a reasoning bridge, the framework translates qualitative human context into interpretable symbolic constraints. Building upon this architecture, we propose a superior drift anticipation framework that ensures enhanced reliability and consistency in Visual SLAM

In the rapidly evolving field of construction management, worker safety remains a top priority. This paper introduces an innovative vision-based system for real-time detection of helmet compliance, specifically designed for construction sites, utilizing advanced computer vision techniques and machine learning algorithms within the YOLO (you only look once) framework. Our system leverages high-resolution video feeds from strategically positioned cameras to monitor adherence to safety regulations regarding helmet usage. By employing deep learning methodologies, the system effectively identifies individuals not wearing helmets, thereby significantly mitigating the risk of head injuries among workers. Our training and validation results revealed an impressive precision exceeding 97% at mAP@0.5 for both helmeted and non-helmeted individuals. Furthermore, our experiments demonstrate exceptional detection accuracy, demonstrating the system's resilience under varying lighting conditions and diverse worker movements. The consistent decrease in loss and improvement in metrics throughout training validates the effectiveness of the YOLOv8 model in enhancing recognition performance. The implications of this research extend beyond mere regulatory compliance, opening avenues for innovative applications in occupational safety management. This study highlights the critical role of technology in protecting lives and lays the groundwork for future advancements in smart construction environments.

Diffusion and VLA policies for manipulation are often deployed through downstream impedance controllers. The stiffness and damping gains of these controllers affect task success, yet are commonly inherited from data collection rather than selected for the deployed policy. Although empirical gain sweeps can improve performance, they require repeated evaluation rollouts. We introduce PHRetune, an offline method that derives controller gains for a frozen policy without evaluation rollouts or gain search. Our approach learns a port-Hamiltonian model from demonstrations to estimate the effort and energy associated with the policy's predicted actions. The policy is applied to recorded demonstration observations, and its predictions are assessed against demonstration-derived effort and energy budgets. From this comparison, we derive a single gain scale in closed form, adjusting the downstream controller while preserving the policy and its action representation. The gains are fixed before evaluation, without requiring a prior manipulator model, task rewards, policy retraining, or additional runtime computation. Across LIBERO suites, PHRetune improves Diffusion Policy success by up to 9.4 percentage points, with the derived gains achieving the highest observed success rates in empirical gain sweeps. On all four real-world manipulation tasks, PHRetuned Diffusion Policy outperforms the nominal policy, alternative gain-tuning methods, and a policy-retraining baseline. The same procedure improves success with SmolVLA and OpenVLA-OFT on every task, while reducing acceleration and jerk for both VLA backbones.

Modern information systems, including many agentic workflows, use dense retrieval to explore large amounts of unstructured data. However, dense retrieval relies on surface-level semantic similarity, which is insufficient for increasingly complex search applications. Here, we investigate agentic retrieval that combines the reasoning capabilities of Large Language Models (LLMs) with the efficient corpus exploration of retrievers in a ReAct agentic loop to solve complex retrieval tasks. In our experiments, we show that agentic retrieval is more effective than standard retrieval, improving nDCG@10 by 8.7 points using the same embedding model. Moreover, while specialized retrieval methods struggle on out-of-domain tasks, agentic retrieval is highly generalizable: the same pipeline achieves competitive results on both the ViDoRe v3 and BRIGHT leaderboards. However, this improvement comes at a cost. On average, agentic retrieval takes 107.4 seconds, compared to 0.67 seconds for standard retrieval, and consumes 764.1K input and 5.8K output tokens per query. In short, our study demonstrates the effectiveness of agentic retrieval in modern data systems and motivates future work on more cost-efficient retrieval agents for large-scale deployment.

This article introduces an innovative system that integrates landmark identification with deep learning to enhance fall detection accuracy and reliability. By utilizing advanced computer vision techniques, such as Media Pipe for spatial recognition, the system effectively differentiates between routine movements and actual falls. The integration of landmarks with a deep learning prediction algorithm minimizes false alarms, ensuring timely responses to genuine falls. Comprehensive experimentation underscores the system's versatility across various scenarios, emphasizing its potential to improve safety and independence for older adults. The training process demonstrates a steady increase in accuracy, stabilizing by the 40th cycle, while error rates decline significantly during the initial cycles. Real-time experiments, involving both male and female participants aged 8 to 50, recorded a remarkable 95% detection rate of falls, demonstrating the system's effectiveness and promising future applications in elder care and smart health monitoring environments.

Large language model (LLM) serving scales its replica count with the request load, yet GPU memory still stands idle inside the replicas. Adding a replica takes minutes, while the memory that a replica needs changes within seconds. Even instant autoscaling could not return this idle memory, because the smallest unit that it can remove is a whole replica. Colocating parameter-efficient fine-tuning (PEFT) with inference can use this memory, but inference must be able to reclaim it within seconds, before requests that wait for memory exceed their latency service-level objective (SLO). Existing colocation systems either keep the tuning memory resident or let inference reclaim it at the coarse granularity of a whole training sample. Each such reclamation also discards the running tuning step. To address these limitations, we present MOLT, a fine-grained memory sharing system that lets inference reclaim the memory of individual activations that a running tuning step has saved for its backward pass. The step continues, and its backward pass recomputes those activations. Inference reclaims only memory that no in-flight GPU work can still access, even under CPU–GPU asynchrony and tensor parallelism. On four model deployments (24B–70B) across H100 SXM and B200 GPUs under trace-driven workloads, MOLT keeps inference SLO attainment at or above 99.7% and completes 1.9–3.3x the tuning work of discard-based memory sharing.

Training-free acceleration of a diffusion policy is accepted when an internal similarity signal reports that the shortcut changed nothing. We price every monitor a control loop can afford in closed-loop success rather than feature distance, over $114$ accelerator configurations and four policy families: two accelerators each clear their own gate's bar and log every reuse as certified, while on one task one finishes every episode and the other none. A monitor is two designs, not one — the statistic it reads, and what it does when that statistic fires. Published gates re-arm after every rejection, and under that response even an oracle handed every forward pass and the exact local action error loses twenty points; absorbing the same statistic costs one point, and most of its speed. What makes a response that never forgets affordable is a statistic that rarely fires, and what it must measure is deviation from the policy the accelerator replaced — which a chunked policy has already paid for, its last plan not yet expired and free to read. Guarding every call this way cuts per-call compute by $1.55$–$3.09\times$, where any reference-requiring check at the same coverage would have to stop accelerating altogether. On three of our four families the schedule alone already holds the pre-stated $\pm2$-point margin. What the monitor is measurably worth shows in three places: on the fourth family, where it rescues the candidate selection landed on; on four configurations it did not select; and on a contact-rich fifth family, chosen where the schedule was expected to fail and run after every design choice was frozen, where no unmonitored arm at its speed holds the margin and the monitored one does.

We propose a learning-to-optimize framework for solving parametric mixed-integer nonlinear programs arising in model predictive control and trajectory optimization. The proposed method combines neural network prediction with a sequential quadratic programming (SQP) refinement layer. First, neural networks are trained using a combination of supervised and self-supervised losses to predict integer solutions together with compatible continuous warm starts. To account for prediction inaccuracies, the trained predictor generates multiple candidate integer solutions, each paired with a continuous initialization. The integer solutions are then fixed, and a differentiable SQP layer refines the associated continuous variables over a prescribed number of iterations. The SQP-layer parameters, including the curvature regularization and step size, are trained to improve both feasibility and objective value after every iteration. By separating integer prediction from continuous optimization-based refinement, the proposed framework improves robustness to imperfect neural predictions while maintaining a predictable computational budget. We demonstrate the proposed framework on contact and trajectory optimization for a planar box-pushing example.

Post-training quantization has been shown to preserve VLA performance under standard evaluation conditions, but whether it preserves the full-precision model's robustness and generalization remains underexplored. In this study, we systematically study the robustness and generalization of post-quantized VLA policies under environmental disturbances. Empirical results show that quantized policies can become fragile to subtle environmental variations despite retaining comparable in-distribution performance. We further observe that action discrepancies are concentrated in a small subset of rollout states, while teacher guidance has opposite effects depending on discrepancy: it improves generalization at high-discrepancy states but can degrade it at low-discrepancy states. These findings reveal that effective post-quantization recovery requires selectively intervening on vulnerable states rather than globally distilling the student. We therefore propose Policy-Induced Vulnerability-Oriented Tuning (PIVOT-Q), a vulnerability-aware On-Policy Distillation (OPD) framework that selectively corrects vulnerable states encountered during quantized-student rollouts using the frozen full-precision policy as a teacher. PIVOT-Q identifies vulnerable states using discounted accumulated discrepancies over a short horizon, applies phase-balanced sparse supervision, and uses a Behavioral Anchor to prevent unnecessary changes. Experiments under seven LIBERO-Plus environmental variations demonstrate consistent recovery across multiple VLA backbones and quantization methods. Notably, PIVOT-Q consistently outperforms full-state distillation across all settings while using only 7.4% of its state-level distillation budget. Our code is available at https://github.com/ruanruan-andy/PIVOT-Q.

Mixture-of-Experts (MoE) has been widely adopted in recent large language model (LLM) architectures. However, scaling up MoE in LLM training introduces system-level challenges on training, where non-uniform token routing can lead to highly imbalanced workloads across experts and devices, further destabilizing the training process. With trillion-scale LLMs, imbalanced expert workloads further amplify the resource cost of MoE training, resulting in degraded training efficiency and hardware utilization for underloaded experts, while hot experts require additional resources to accommodate excessive workloads. Recent studies address imbalanced MoE training through intricate parallelism strategies or resource reallocation. However, these system-level approaches often introduce additional resource requirements and considerable orchestration complexity, which become increasingly difficult to afford when training trillion-parameter LLMs under constrained computational resources. This work introduces CIPHER-MoE, which mitigates MoE workload imbalance while keeping the router's token-side Top-K selection unchanged. CIPHER-MoE applies affinity-aware Expert-to-Token filtering with explicit capacity control to reduce hotspot expert workloads without additional hardware resources or complex runtime design. The proposed method has been evaluated on large-scale MoE models, including DeepSeek-V4-Pro, showing up to 64.9 percentage points Top-1 expert workload reduction and 1.10$\times$-1.94$\times$ training acceleration, while preserving the training quality. The source code will be released soon.

RANSAC scoring and local optimization (LO) impose different robustness requirements, motivating the separation of hypothesis selection from refinement. We systematically isolate the effects of robust-loss shape, incorrectly specified inlier scales, and optimization strategy on essential matrix, fundamental matrix, and homography estimation. A profile-marginal score marginalizes the nuisance inlier scale and selects an inlier partition, from which we estimate the scale that sets the LO loss width; this makes LO robust to an inlier scale specified too large, whereas one specified too small degrades selection itself. Refinement needs gradient from correspondences the seed currently rejects: optimizers that reweight from current residuals stay pinned to their seed, whereas methods with broad basins recover strongly perturbed seeds yet degrade accurate score-selected hypotheses, so basin size alone is insufficient to assess RANSAC LO. Joint half-quadratic optimization balances the two and is the most consistent strategy across model classes. An optimizer matched to the profile-marginal score, which never decreases it, does not reach the best accuracy, challenging the prescription that scoring and refinement objectives should match. Composed from these findings, our RANSAC reduces the median essential-matrix pose error of a state-of-the-art RANSAC on PhotoTourism from 2.23 degrees to 1.58 degrees with a correctly specified inlier scale and from 38 degrees to 6.2 degrees when it is grossly misspecified (128x too large).

Long-horizon video world models require persistent memory to preserve scene consistency over extended rollouts. Softmax attention retains the full generation history through a growing KV cache, whereas recurrent linear attention compresses history into fixed-size states with substantially lower memory cost. However, we identify severe long-range forgetting in Gated DeltaNet (GDN), where information from distant but relevant scenes is progressively attenuated by subsequent state updates. To address this limitation, we propose HLA-WM, a training-free hybrid linear-attention framework that combines coarse-grained geometry-guided retrieval with fine-grained recurrent linear-state computation. HLA-WM exploits the affine structure of GDN to cache compact chunk-wise transition summaries, retrieve scene-relevant historical chunks using camera geometry, and recompose them into query-specific recurrent states. On the $60$-second SANA-WM-Bench, HLA-WM improves all six aggregate revisit-consistency and camera-control metrics of the base autoregressive generator without additional training, including a $0.74$ dB PSNR gain and a $28.5\%$ reduction in rotation error. The improvements persist after downstream refinement and generalize to MBench-A, where HLA-WM consistently improves all three revisit-consistency metrics across all four subsets and all evaluated inference modes over $547$ samples. At a $60$-second context, HLA-WM reduces historical-state memory by $12\times$ relative to full KV caching while incurring at most a $1.6\%$ reduction in inference throughput. These results demonstrate that selectively addressable recurrent memory can improve long-range scene recall while preserving the efficiency advantages of GDN. Project page: https://caesarhhh.github.io/hla-wm/

This paper revisits frame-wise saliency for audio moment retrieval (AMR). We show that the frame-wise saliency sequence, conventionally used only as an auxiliary output in DETR-based AMR models, can itself serve as an effective source of moment predictions. We convert the saliency sequence into ranked moments using a simple SED-inspired segmentation rule with no learned parameters, enabling moment retrieval directly from frame-wise temporal information. On the CASTELLA dataset, saliency-based prediction consistently outperforms decoder-based prediction from the same model across all 18 runs of QD-DETR and CG-DETR. For QD-DETR, simply replacing the inference output improves R1@0.7 from 21.0 to 36.1. The advantage remains 7-15 points when the two outputs are evaluated at their independently selected best epochs. The same tendency extends to TaskWeave and UVCOM, whereas TR-DETR shows the opposite behavior, suggesting that how saliency construction may matter. The performance gap is especially pronounced for short moments: for queries whose annotated moments average at most 2 s, R1@0.7 improves from 6.3 to 27.8 with QD-DETR. Decoder supervision nevertheless benefits saliency-based prediction, indicating that its role during training differs from the utility of its inference output.

Online control of execution speed is essential for deploying robot policies in real-world scenarios, as robots may need to speed up under time constraints or slow down to facilitate human interaction and improve safety. However, imitation-learned policies inherit the execution speed of their demonstrations, and test-time speed modification can introduce unrecoverable out-of-distribution observations, reducing task success. We observe that the directional inconsistency of action chunks reflects task-phase criticality, indicating how aggressively action step lengths can be modified while preserving task success. Based on this observation, we introduce FreeSpeed, a training-free module that post-processes action chunks from pretrained policies. FreeSpeed resamples each predicted chunk at the requested rate, then uses directional inconsistency between adjacent actions as the primary signal for rescaling. This signal adaptively determines how closely the execution speed can approach the requested speed, allowing flexible speed adjustment within the evaluated limits without compromising task success. Across three policy families and 50 simulated tasks, FreeSpeed supports online speed changes, with realized execution rates spanning 0.22x to 2.53x among settings that preserve per-task success. Across four real-world manipulation tasks, FreeSpeed achieves an average success rate of 94.0%, matching the frozen policy's 93.8%, while realizing execution rates from 0.38x to 1.97x.

Autonomous social navigation requires balancing efficiency, physical safety, and social compliance. Reinforcement Learning (RL) methods provide a viable and effective solution but often rely on unrealistic assumptions, such as the knowledge of humans' position and velocity. In this paper, we introduce JESSI (JAX-based E2E Safe Social Interpretable navigation), a lightweight end-to-end RL framework that maps raw LiDAR scans directly to kinematically feasible control commands. JESSI enhances safety via Dirichlet-parameterized continuous action spaces and deterministic bounding, while an integrated attention-based perception module extracts probabilistic human states for interpretable, socially aware decision-making. Through extensive simulations and real-world deployment on a differential-drive robot, we demonstrate that jointly optimizing the RL policy with a supervised perception signal in a multi-task paradigm enhances social behavior. Ultimately, JESSI is able to balance high navigation success rates and superior social behaviors compared to state-of-the-art baselines.

LLM agents rely on long-term memory to retain and reuse information when performing tasks over long horizons. Existing methods provide limited support for handling memories that become outdated as new observations or domain evidence arrive. Such outdated memories may remain semantically relevant, continue to affect dependent records, and retain value as historical evidence. This calls for two capabilities: dependency tracking to identify downstream effects and historical preservation to retain useful past records. We propose Provenance-Aware Cascading Memory Invalidation (PACMI), a framework that represents memories and new evidence in a provenance graph with typed dependency edges. PACMI assigns records to a four-state validity lattice, propagates validity changes to dependent memories, and uses the resulting states for retrieval and stale-premise detection. We also introduce a diagnostic benchmark with 100 cases and 300 queries across five domains. The evaluation separates node, context-, and answer-level performance. PACMI achieves the highest final-answer accuracy on this benchmark, and its paired difference from the strongest baseline is significant under an exact McNemar test. The premise checker achieves perfect precision, recall, and F 1 on the controlled query distribution. Cascading propagation primarily improves memorystate correctness: removing it increases final-answer errors from 3 to 11, but the paired difference does not reach the 0.05 significance threshold. Code and data will be made publicly available.

Earth observation (EO) data provide rich temporal supervision, yet existing remote sensing foundation models mainly exploit sequential observations through imposing predefined pairwise relations or aggregating holistic reconstruction context. We seek to further exploit the sparse and nonuniform temporal sampling inherent in EO sequences as supervisory signals. To this end, we propose T-JEPA, a temporal joint-embedding predictive architecture that learns time-gap-conditioned latent transitions. A shared single-frame encoder processes each observation, while a temporal predictor estimates the complete target latent field from a masked source latent representation and the actual elapsed time. Across multiple temporal intervals, these predictive constraints organize observed states into structured latent trajectories. Asymmetric metadata injection mitigates shortcut learning, and direct supervision across multiple temporal scales proves more effective than recursively rolling out intermediate states. In parallel, masked pixel reconstruction provides complementary supervision for preserving spatial details. Under matched pre-training data and throughput, T-JEPA achieves leading transfer performance on both static and temporal tasks. Analyses further reveal that T-JEPA learns representations with time-gap-dependent transition predictability and coherent latent dynamics, while maintaining strong cross-period consistency, representation diversity, and semantic discriminability.

Mean-field games provide a continuum framework for modeling the dynamics of large, interacting populations of non-cooperative agents. This paper studies symmetry-breaking in a finite-horizon, two-choice min-LQG mean-field game in which identical agents with linear stochastic dynamics choose one of two equally desirable terminal destinations, while trading off control effort against social pressure to conform. The model has an odd symmetry between the two destinations and therefore always admits a symmetric, dynamic Nash equilibrium in which the population splits evenly between them, producing a deadlock collective state at final time. Numerical studies have suggested that, as the penalty for social nonconformity increases, this symmetric equilibrium loses stability as a fixed point of an associated scalar, self-consistency map, and asymmetric consensus Nash equilibria emerge, where most agents select the same destination. By analyzing the linearized forward-backward PDE system through the scalar map representation, we provide a proof of this loss of stability of the symmetric equilibrium. Together with the odd symmetry of the map, this implies the existence of symmetry-broken mean-field game equilibria corresponding to consensus on either destination.

Generative AI (GenAI) can explore large bodies of literature and generate plausible research ideas, but identifying what is missing, understudied, contradictory, or potentially connected does not by itself reveal where theory should advance. We develop a theory-oriented agentic AI system that helps researchers identify potential theorizing opportunities by incorporating established theorizing approaches into literature exploration and evaluation. The system operates through three stages. Stage 1 expands the theoretical search space and constructs a provisional Candidate Knowledge Graph. Stage 2 independently reconstructs what the literature supports through source grounded evidence extraction and theory-state reconstruction. Stage 3 evaluates the reconstructed knowledge state to determine whether an unresolved configuration warrants theory development or another research action and, when theory development is warranted, which theorizing approach is appropriate. We demonstrate the system through an end-to-end analysis of human oversight of agentic AI systems in organizations. The analysis shows that literature gaps alone are insufficient for identifying theoretical opportunities. For example, "transparency to trust" is routed to mechanism-based theorizing because the relationship is repeatedly documented in prior studies, while the generative mechanism explaining how transparency shapes trust remains insufficiently specified. By combining large-scale literature processing, structured knowledge representation, and theorizing-guided diagnosis, the system serves as a theory-oriented research assistant that supports researchers in identifying theoretically meaningful directions for subsequent research.

Large Language Models (LLMs) have shown strong potential in multivariate time series forecasting and anomaly detection. Existing studies predominantly inject temporal information into LLMs via direct numerical tokenization or heuristic textual descriptions. However, LLMs still face difficulty in perceiving the underlying structural patterns of numerical time series, particularly the seasonal and trend components obscured by discrete numerical tokens. To bridge this gap, we propose FreSia, a frequency-aware framework that establishes an effective alignment between the semantic space of LLMs and the frequency space of time series. Specifically, FGPrompt, a Frequency-Guided Prompt mechanism within FreSia, distills the frequency-domain structures of time series and projects them into prompts tailored to the semantic space of LLMs. Furthermore, we introduce a Global-driven Context Learning (GCL) component, which uses a global CLS-driven probe to generate global context to bridge the time-frequency domain gap and fuse the multi-modal information. Experiments on eight forecasting benchmarks show that FreSia achieves average improvements of 13.48% and 8.06% in MSE and MAE, respectively.

This paper studies data-driven stabilization of unknown discrete-time linear systems using only quantized state measurements that take values in finite sets. The proposed approach consists of two stages. In the controller design stage, we collect quantized data while maintaining a prescribed quantization error bound and use them to formulate a semidefinite program (SDP). We establish a verifiable condition under which any feasible solution to the SDP yields a stabilizing feedback gain, and show that a stabilizing gain can always be obtained when the quantization error is sufficiently small. In the stabilization stage, a Lyapunov-based quantizer update rule is developed to guarantee exponential convergence under quantized state feedback. As a key feature, the number of quantization cells remains finite in both stages and constant during stabilization. Simulation results illustrate the effectiveness of the proposed approach.

Given incomplete measurements of a single physical field in a coupled system with unknown parameters, can we infer its full physical state and identify the underlying parameters? This problem is challenging because multiple coupled fields must be reconstructed simultaneously from limited observations of only one, while the system parameters are unknown. In this work, we propose a machine learning framework for full-field reconstruction and parameter identification of unknown physical systems from sparse observations of a single physical field. Specifically, the cross-attention encoder propagates sparse sensor observations onto a regular grid to construct a sensor-conditioned latent representation, while a Fourier neural operator (FNO) decoder captures global spatial dependencies to reconstruct all coupled physical fields. The network parameters and unknown physical parameters are jointly optimized by minimizing observation losses, governing equation residuals, and boundary/initial condition constraints. The proposed approach is validated on two- and three-dimensional lid-driven cavity flows, a two-dimensional cylinder wake, and a two-dimensional non-ideal magnetohydrodynamics problem, demonstrating the recovery performance of unobserved fields and physical parameters from incomplete observations.

Vision-Language-Action (VLA) models serve as unified policies for robotic manipulation, yet their expensive inference forces robots to pause between policy calls, resulting in stop-and-go execution that interrupts smooth motion and prolongs task completion. Extending the action chunk reduces policy calls and hence these pauses, but predicting farther into the future makes long-chunk execution unreliable. To understand where this unreliability arises, we analyze action errors within long chunks and find that they concentrate around transitions between manipulation subskills, growing sharply with chunk length. This suggests the importance of transition timing, i.e., when to switch subskills within a chunk. Motivated by this observation, we introduce RACE (Reliable Action-Chunk Extension), a framework that predicts the transition timing from an auxiliary one-step denoising pass and conditions action generation on it. By learning and conditioning on transition timing, RACE reduces errors at subskill transitions and enables reliable execution of longer action chunks. Across simulation benchmarks, RACE outperforms fine-tuning at the same chunk length; with 2x longer chunks, it surpasses recent state-of-the-art and efficient VLAs in success rate, and with 4x longer chunks, it remains competitive. On a real robot, RACE uses 4x longer chunks, which reduces the idle time caused by stop-and-go execution by about 5x, while achieving a higher success rate than fine-tuning with the same chunk length. Code and a real-robot demo are available at https://github.com/Seonghoon-Yu/RACE-VLA

We introduce Reduced-rank Mahalanobis Distance (ReMaD), a novel prototypical distance-based refinement to classification and out-of-distribution (OOD) detection using pretrained models without finetuning. We use embeddings of the target dataset to fit closed-form distribution statistics in the model's latent space which can classify in-distribution samples and detect OOD samples, all without training or prior knowledge of the OOD data. Building on prototype classification and OOD detection, we analyze the distribution properties of large pretrained models when processing new datasets; based on this analysis, we formulate a simple modification to Mahalanobis Distance to adapt models' latent space distributions to new domains by removing unused features, without the finetuning or hyperparameter searches required by other adaptation procedures. We demonstrate the efficacy of this method to adapt existing large pretrained image embedding models to new classification domains outside their trained capabilities by testing across four target datasets, with competitive performance in both classification and OOD detection.

In-context retrieval (ICR) is a retrieval-based form of in-context learning (ICL) in which demonstrations are retrieved from a source database based on similarity to the query, rather than sampled independently. In this work, we formulate ICR as a type of domain adaptation problem, where the source distribution $P$ of the database may differ from the target distribution $Q$ of the test query-label pair. We investigate the performance of ICR under a flexible class of distributional shifts that substantially extends prior work \citep{li2024fine,guo2025retrieval}, and establish theoretical guarantees that quantify the benefits and pitfalls of this learning paradigm. Our theory is verified by experiments on synthetic and language tasks.

Vision-language models (VLMs) can detect that an object has rotated across views, but cannot reliably tell by how much. We introduce OR-Bench, a fine-grained benchmark for object-rotation reasoning with eight tasks covering rotation detection, rotation magnitude estimation, and multi-view rotation reasoning. Across 12 VLMs, the gap is stark: the strongest models approach 100% accuracy on detection, yet even coarse magnitude estimation is near chance. When asked for exact angles, models place 91.8–100% of their predictions on just $0^\circ$, $90^\circ$, and $180^\circ$, a failure we term canonical-angle collapse. This collapse persists even without visual input. Representation probing shows that missing information is only part of the explanation. Although rotation information becomes less recoverable at finer granularity, substantial coarse-grained information remains, and a simple linear probe outperforms the models' generated answers. This suggests that VLMs underuse rotation information they already encode. We therefore propose RotationCue, a lightweight decoder that recovers coarse rotation information from the VLM's own frozen representations and feeds it back to the model as intermediate textual context. Across three VLMs, RotationCue improves every model–task combination on OR-Bench, raising macro-average accuracy by 7.9–12.6 points while preserving general capabilities.

We introduce a framework for multi-bit text watermarking with security defined directly through $f$-divergence from the base language model distribution. Unlike prior approaches that focus on average-key distortion-freeness or a particular statistical distance, our formulation supports general $f$-divergences, including total variation and KL divergence, and enforces the guarantee for each realized key and embedded message. We develop a coding-based watermarking scheme that optimally biases next-token distributions subject to a prescribed divergence budget, and characterize the resulting tradeoff between embedding rate, decoding reliability, and statistical security. Experimentally, we compare our method against prior multi-bit watermarking schemes across modern language models and payload regimes. Our approach achieves substantially lower watermark detectability while maintaining competitive message-recovery performance and generation quality. Our results provide a unified view of secure multi-bit watermarking and recover several commonly used security notions as special cases.