arXiv subjects

Li Li

Publications and source records attributed to Li Li.

At least 73 records · Page 4Linked to original sources

arXiv2026

DRBA: Dynamic Robotic Balance Assistant -- An assist-as-needed gait and balance rehabilitation robot for versatile training

The decline of human balance control due to aging and pathological conditions increases fall risk, a major concern in geriatric care and rehabilitation. Gait training is essential for balance recovery, enhancing walking ability and postural control. However, existing overground robotic gait trainers have limitations: body weight support systems are bulky and impractical for daily use, while end-effector-based systems often compromise transparency, altering natural gait dynamics. This paper presents the Dynamic Robotic Balance Assistant (DRBA), a novel gait trainer providing assist-as-needed body weight and balance support for various training scenarios. DRBA integrates a 3-degree-of-freedom (3-DoF) robotic arm for pelvic support with flexible motion, a compact sit-to-stand assistance module, and user-following and fall detection algorithms to ensure minimal interference and responsive support. Experimental results demonstrated high transparency, with minimal impact on natural gait dynamics. A patient trial with nine elderly patients with varying medical conditions and balance impairments (ranging from severe to mild) further validated DRBA's effectiveness. The results showed that DRBA-assisted training increased step length and walking speed compared to therapist-assisted gait training. Additionally, DRBA enabled users to perform tasks beyond their unaided ability, expanding rehabilitation possibilities. These findings highlight DRBA's potential to enhance rehabilitation outcomes by facilitating higher training intensity and enabling task-oriented exercises.

cs.RO↗

arXiv2026

AtomicCommitBench: Can Coding Agents Reconstruct Commit Histories from Squashed Patches?

Coding agents often finish a session by returning one squashed patch that mixes feature implementation, bug fixes, refactorings, tests, and configuration edits. While the final code may be correct, collapsing unrelated edits into one patch removes the history structure needed for review, selective revert, and later maintenance. We study retrospective commit-history reconstruction: given a completed squashed change, an agent groups its hunks into commits and materializes a replayable commit sequence. We formalize the task as hunk-to-commit partitioning with a replay requirement and build AtomicCommitBench, containing 800 real consecutive-commit episodes from 10 Python projects. Because multiple decompositions may be reasonable, we evaluate outputs using complementary metrics: PPAR for replay validity, ARI for reference-based grouping quality, and TCR for failure containment on scoreable modified-test episodes. Natural retrospective reconstruction proves substantially harder than replay checking or synthetic tangling. Although nearly all non-random methods achieve replay validity (PPAR >= 0.988), grouping quality ranges from 0.03 to 0.46 ARI. Matched synthetic composites are much easier than real same-author squashed diffs (+0.333 ARI). In our evaluation, the GPT-5.4 setup driven by Codex CLI (0.46 ARI) and the GLM-5 setup driven by Claude Code (0.43 ARI) outperform MiniMax (0.31) and Kimi (0.29). Qualitative analysis identifies same-file lumping and support-hunk drift as recurring failure modes. Dependency-Aware Commit Evidence (DACE) improves the lower-scoring setups by 0.05 to 0.08 ARI, indicating that dependency cues and hunk-role information help agents avoid locality-driven grouping errors. AtomicCommitBench enables evaluation of the commit histories produced by coding agents alongside the final code.

cs.SE↗

arXiv2026

Real-Time Neural Video Compression with Unified Intra and Inter Coding

Neural video compression (NVC) technologies have advanced rapidly in recent years, yielding state-of-the-art schemes such as DCVC-RT that offer superior compression efficiency to H.266/VVC and real-time encoding/decoding capabilities. Nonetheless, existing NVC schemes have several limitations, including inefficiency in dealing with disocclusion and new content, interframe error propagation and accumulation, among others. To eliminate these limitations, we borrow the idea from classic video coding schemes, which allow intra coding within inter-coded frames. With the intra coding tool enabled, disocclusion and new content are properly handled, and interframe error propagation is naturally intercepted without the need for manual refresh mechanisms. We present an NVC framework with unified intra and inter coding, where every frame is processed by a single model that is trained to perform intra/inter coding adaptively. Moreover, we propose a simultaneous two-frame compression design to exploit interframe redundancy not only forwardly but also backwardly. Experimental results show that our scheme outperforms DCVC-RT by an average of 12.1% BD-rate reduction, delivers more stable bitrate and quality per frame, and retains real-time encoding/decoding performances. Code and models will be released.

cs.CV↗

arXiv2026

How Much Static Structure Do Code Agents Need? A Study of Deterministic Anchoring

LLM-based code agents navigate repositories through keyword search but miss the structural relationships, such as call graphs, inheritance hierarchies, and configuration dependencies, that define how software actually works. This makes agent navigation stochastic and difficult to reproduce across runs. We investigate whether lightweight static analysis can provide deterministic anchors for these agents: stable structural facts injected as plain-text comments that constrain probabilistic exploration and make navigation more predictable. Starting from a strong baseline, Codex from OpenAI, we systematically inject varying granularities of structural annotations and measure their effects on localization, trajectory behavior, and run-to-run stability. Our study identifies what we call the deterministic anchoring effect: static structure helps less by making agents "smarter" and more by making their navigation disciplined and reproducible. Three observations support this finding: (1) Anchoring works: lightweight call/inheritance topology improves function-level localization (+2.2pp Func@5) and shortens trajectories (-1.6 interaction rounds); (2) Anchoring is scale-sensitive: the optimal granularity and directionality depend on repository characteristics, where denser semantics show diminishing returns and hub-heavy projects benefit from inverse-only links that expose "who-calls-me" without forward edges; (3) Anchoring stabilizes: tags raise link-following rate from 0.15-0.18 to 0.21-0.24, roughly halve run-to-run variance, and improve single-run reliability (Pass@1 +3.4 pp) on medium-scale repositories, at the cost of roughly 10% more input tokens. These observations suggest practical guidelines: default to lightweight topology on medium projects, prune forward edges in large repositories, and reserve dense tags for implicit-dependency cases.

cs.SE↗

arXiv2026

RepoRescue: An Empirical Study of LLM Agents on Whole-Repository Compatibility Rescue

Open-source libraries and tools are widely reused, but compatibility maintenance is expensive. Once maintainers leave, useful repositories can stop working as runtimes and dependencies evolve. We study whether LLM agents can adapt old repositories to modern environments, a task we call compatibility rescue. Unlike bug repair, compatibility rescue starts from a repository that worked in its original environment but fails after ecosystem drift. RepoRescue gives agents only the repository and its failing modern environment; the agent must diagnose the failure, locate affected code, and produce a source-code rescue that restores the historical test suite. We build RepoRescue from 193 Python and 122 Java repositories, each verified to pass historically and fail after modernization. We evaluate five deployed agent systems on Python and three on Java. Beyond full-patch pass rate, we rerun patches after removing test-file edits to measure source-only repair, add a runtime-enforced regime that blocks test edits, and validate practical use for repositories whose suites pass after rescue. We find that Claude Code systems sometimes edit failing tests even when prompted not to; with runtime blocking, Kimi still rescues 41.5% of repositories. Systems are complementary: their union reaches 62.7%, exceeding the best single system by 10.9 points. Difficulty concentrates in cross-file coordination: on 14 repositories requiring coordinated whole-codebase changes, GPT-5.2 through Codex passes all 14, while every Claude Code system passes at most two. Finally, a passing suite is only an initial signal: among 34 unmaintained Python candidates whose suites pass after rescue, 22 work in realistic scenarios and 12 pass bug-hunt with patches that address the compatibility failure. RepoRescue benchmarks compatibility rescue with source-only auditing, runtime enforcement, practical validation, and reasoning labels.

cs.SE↗

arXiv2026

Accelerating Merge with Motion Vector Difference via Filter Difference Analysis for VVenC

Merge with Motion Vector Difference (MMVD) is a key coding tool in Versatile Video Coding for improving motion prediction accuracy. However, its exhaustive search strategy imposes a significant computational burden on the encoder. To address this issue, we propose a novel fast MMVD algorithm for the VVenC encoder based on fractional motion vector filter difference analysis. By approximating the 8-tap interpolation filter with a 2-tap filter, we derive a criterion based on spatial gradients and prediction residuals for estimating the potential gain of MMVD candidates. We further generalize this criterion to accommodate both shifted integer reference samples and 2D separable filtering. To minimize the overhead of the proposed method, we introduce implementation optimizations, including symmetric offset inference and cross-shaped downsampled dot-product computation. Compared with existing fast MMVD algorithms in VVenC, our method reduces the average MMVD search ratio from 21.07\% to 11.05\% and decreases the efficiency-complexity metric $η$ from 11.79 to 7.10 under the fast preset.

eess.IV↗

arXiv2026

To Run or Not to Run: Analyzing the Cost-Effectiveness of Code Execution in LLM-Based Program Repair

LLM-based agents for program repair are increasingly built on a "generate-run-revise" paradigm, iteratively executing tests to evaluate and refine patches. This execution-based approach has become standard practice in state-of-the-art systems. However, executions can be time-consuming and expensive, yet their impact on these agents remains underexplored. In this paper, we conduct a two-stage empirical study over execution behavior in LLM-based program repair. To characterize execution behavior at scale, we first analyze 7,745 agent traces from SWE-bench leaderboard submissions. Second, we evaluate 3,000 end-to-end repair attempts across 200 SWE-bench instances and three agents (Claude Code, Codex, and the open-source OpenCode) under four execution paradigms, which allows for a fine-grained comparison of performance and cost. Our analysis reveals three key observations: (1) Code execution is used across all agents and models analyzed, with an average of 8.8 test runs per task. Execution behavior varies substantially across agents and models, with frequency ranging from 2 to 19 per task, and late-stage executions consistently achieve higher success rates than early-stage ones. (2) Execution restrictions have little effect on repair success: on commercial agents with SOTA models the resolve-rate gap between Prohibited and Unrestricted is only 1.25 percentage points and not statistically significant, while Prohibited saves substantial token and wall-clock cost. (3) Execution benefit is concentrated rather than uniform. These patterns suggest that current agents apply execution indiscriminately, paying its cost on instances where it provides little benefit. Execution, therefore, should be treated as a resource with an explicit cost-benefit tradeoff, not a default capability.

cs.SE↗

arXiv2026

Bias Fitting to Mitigate Length Bias of Reward Model in RLHF

Reinforcement Learning from Human Feedback (RLHF) relies on reward models to align large language models with human preferences. However, RLHF often suffers from reward hacking, wherein policy learning exploits flaws in the trained reward model to maximize reward scores without genuinely aligning with human preferences. A significant example of such reward hacking is length bias, where reward models usually favor longer responses irrespective of actual response quality. Previous works on tackling length bias have notable limitations, these approaches either mitigate bias without characterizing the bias form, or simply assume a linear length-reward relation. To accurately model the intricate nature of length bias and facilitate more effective bias mitigation, we propose FiMi-RM (Bias Fitting to Mitigate Length Bias of Reward Model), a framework that autonomously learns and corrects underlying bias patterns. Our approach consists of three stages: First, we warm up by training a standard reward model which inherently contains length bias. Next, we deploy a lightweight fitting model to capture the non-linear relation between length and reward. Finally, we incorporate this learned relation into the reward model, effectively decoupling length from reward while preserving preference modeling capabilities. Experimental results demonstrate that FiMi-RM achieves a more balanced length-reward distribution. Furthermore, when applied to alignment algorithms such as Direct Preference Optimization (DPO) and Best-of-N (BoN), our debiased reward model improves length-controlled win rate and reduces verbosity without compromising its performance.

cs.LG↗

arXiv2026

Executing as You Generate: Hiding Execution Latency in LLM Code Interpreters

Current LLM systems are increasingly equipped with a code interpreter that executes generated code to obtain results. This works serially: the model first generates the complete code, then an interpreter executes it. This sequential workflow leaves the executor idle during generation and the generator idle during execution, resulting in unnecessary end-to-end latency. Our key observation is that an LLM, unlike a human developer, emits code tokens left to right and does not backtrack over what it has already written. This makes it possible to start executing a piece of code while later tokens are still being generated. We formalize this parallel execution paradigm, modeling it as a three-stage pipeline of generation, detection, and execution, and derive closed-form latency bounds that characterize its speedup potential and operating regimes. We then present EAGER, a concrete implementation featuring AST-based chunking, dynamic batching with gated execution, and early error interruption. We evaluate EAGER across four benchmarks, seven LLMs, and three execution environments. The overlap mechanism hides almost all execution behind generation, reducing the non-overlapped portion of execution time by up to 99.8% and cutting end-to-end latency by up to 37.3% on error-free runs.

cs.PL↗

arXiv2026

RaMem: Contextual Reinstatement for Long-term Agentic Memory

Long-term memory has become increasingly important for LLM agents that operate across extended interactions and evolving task contexts. Recent memory systems have made past experiences more persistent, compact, and retrievable, but retrieval alone does not ensure that a memory provides valid evidence for the current query. When experiences are compressed into reusable fragments, memories from different situations may appear equally relevant if they involve recurring entities or user states. We refer to this failure as context collapse: memories lose the surrounding context needed to judge whether they provide valid evidence for the current query. To address this problem, we propose Contextual Reinstatement for Agentic Memory (RaMem), a framework that turns retrieved memory fragments into contextually verifiable evidence. RaMem operates through four coordinated stages: (i) evidence anchoring grounds each memory in its original episodic conditions, especially event time, mention time, session span, and participants; (ii) recall condition induction derives the evidence conditions implied by the query; (iii) validity-aware retrieval uses these conditions to prioritize context-compatible memories while retaining content-relevant candidates as fallback evidence; and (iv) context-preserved synthesis keeps the selected memories' structured context available to the generator. Experiments on long-term memory benchmarks show that RaMem consistently improves performance over strong memory baselines, with average F1 gains of more than 10% across several backbones.

cs.AI↗

arXiv2026

Awareness of Technological Isomorphism: Integrating AI into Elementary Mathematics Teaching on Data and Prediction,A Case Study of the Compound Line Graph

The deep integration of Artificial Intelligence (AI) into elementary mathematics education necessitates a conceptual tool capable of explaining students' cognitive transition from disciplinary knowledge to AI understanding. This study proposes a novel core concept, "Awareness of Technological Isomorphism, " defined as a student's metacognitive realization that their own mathematical cognitive operations (e.g., observing trends, inducing patterns, and making predictions) share an underlying logical structure with AI technical operations (e.g., pattern recognition and predictive modeling). This awareness, in turn, facilitates cognitive transfer from disciplinary mathematics to AI comprehension. Underpinned by transfer learning and metacognitive theories, this study clarifies the distinct essence of this concept from traditional "computational thinking." We demonstrate the explanatory power of this framework in two ways: elucidating the mechanism of students' cognitive leap from mathematics to AI, and guiding instructors to identify "isomorphic interfaces" within disciplinary curricula. On this basis, a three-stage pedagogical pathway--spanning "Perception, Comprehension, and Creation"--is constructed alongside a corresponding evaluation rubric. This framework is empirically validated through a case study based on the "Compound Line Graph" lesson from a fifth-grade mathematics textbook in China, offering a highly replicable operational framework for the deep convergence of disciplinary instruction and AI literacy education.

cs.CY↗

arXiv2026

Learning from Human Driving: A Human-in-the-Loop Online Behavior Cloning Framework for Autonomous Driving

With the evolution of large foundation models (LFMs), data-driven autonomous driving has made significant strides. However, existing paradigms still face severe challenges in complex interaction and long-tail scenarios due to distribution shift and causal confusion. These limitations often result in a lack of human-level decision-making flexibility and safety in extreme conditions. To overcome this limitation, this paper proposes a Human-in-the-Loop Online Behavior Cloning frame work (HiL-OBC) for autonomous driving, which aims to deeply integrate the cross-modal perceptual capabilities of LFMs with the high-level driving intelligence of human experts. Specifically, HiL-OBC deployment is executed through three critical phases: policy initialization with human intervention, latent behavioral modeling with Bayesian policy adaptation, and online deploy ment and updates. Furthermore, we design a Multi-modal Online Behavior Cloning (MOBC) model, which optimizes the base driving policy online through a lightweight network architecture, a takeover trigger mechanism, and a multi-variant loss function, thereby enhancing the system's decision-making robustness in complex environments. We evaluated the HiL-OBC on the LangAuto-Human CARLA benchmark. Experimental results demonstrate that the driving policies optimized via the human-in-the-loop mechanism achieve substantial performance gains: the DS of StructNav, LFG, and LMDrive increased by 47.25%, 31.59%, and 32.12%, respectively, with a simultaneous of various experimental settings and key components highlights the advantages of human-in-the-loop learning in improving decision-making robustness and overall driving performance.

cs.RO↗

arXiv2026

OpenACMv2: An Accuracy-Constrained Co-Optimization Framework for Approximate DCiM

Digital Compute-in-Memory (DCiM) accelerates neural networks by reducing data movement. Approximate DCiM can further improve power-performance-area (PPA), but demands accuracy-constrained co-optimization across coupled architecture and transistor-level choices. Building on OpenYield, we introduce Accuracy-Constrained Co-Optimization (ACCO) and present OpenACMv2, an open framework that operationalizes ACCO via two-level optimization: (1) accuracy-constrained architecture search of compressor combinations and SRAM macro parameters, driven by a fast GNN-based surrogate for PPA and error; and (2) variation- and PVT-aware transistor sizing for standard cells and SRAM bitcells using Monte Carlo. By decoupling ACCO into architecture-level exploration and circuit-level sizing, OpenACMv2 integrates classic single- and multi-objective optimizers to deliver strong PPA-accuracy tradeoffs and robust convergence. The workflow is compatible with FreePDK45 and OpenROAD, supporting reproducible evaluation and easy adoption. Experiments show that the proposed two-level ACCO framework achieves most of its accuracy-constrained efficiency gain at Level-1 through architecture exploration, delivering roughly 50%+ PDP reduction, while Level-2 transistor-level optimization provides a further single-digit PDP improvement while preserving accuracy, enabling rapid "what-if" exploration for approximate DCiM. The framework is available on GitHub (https://github.com/ShenShan123/OpenACM).

cs.LG↗

arXiv2026

Programmable spectral symmetries in an anisotropic quantum Rabi simulator

The quantum Rabi model captures fundamental aspects of light--matter interaction, where symmetry dictates both spectra and dynamics. Over the past years, experiments have explored many of its nonperturbative properties, but have mostly focused on the isotropic limit, where rotating and counterrotating processes are locked together, leaving the broader symmetry landscape largely unexplored. Here we realize a programmable anisotropic quantum Rabi model in a superconducting processor, with independent control of the rotating and counterrotating couplings $(g_1,g_2)$ and of a transverse bias $\varepsilon$. Continuous anisotropy tuning, combined with a duality mapping, gives access to the full parameter space from the Jaynes-Cummings to the anti-Jaynes-Cummings limits. In the deep-strong-coupling regime, we show that anisotropy reconstructs the spectrum and turns complete collapse-revival dynamics into incomplete revivals even near degeneracy. With adiabatic state preparation and joint tomography, we resolve an anisotropy-induced ground-state parity switch, a crossing that has no analogue in the isotropic model. We further observe selective tunnelling associated with hidden symmetry in biased Rabi models and track its anisotropic displacement within the same device. These results establish a controllable route to engineering nonperturbative light--matter Hamiltonians, where symmetry, spectrum, and dynamics can be programmed independently.

quant-ph↗

arXiv2026

Memory Retrieval for Changing Preferences

Long-context dialogue systems must decide both when to access memory and which parts of the interaction history are relevant. Existing approaches typically rely on heuristic retrieval signals or always-on memory usage, failing to account for the changing and potentially inconsistent nature of user preferences. In this work, we propose a unified framework for memory access and selection based on changing preferences. We formulate personalized memory retrieval as identifying which historical turns provide evidence about a user's latent preference state, rather than relying on surface-level semantic similarity. To this end, we quantify the utility of each memory turn using a Bayes factor, defined as the improvement in the model's likelihood of the reference response when the turn is included in context. This provides a principled measure of evidence strength and a unified signal for both memory access and selection. By framing memory retrieval as utility estimation, the model learns to identify salient turns and regulate memory usage based on expected utility. Experiments on four heterogeneous memory benchmarks show that our approach outperforms existing embedding-based retrieval on long-context, preference-intensive tasks where modeling changing preferences is essential, while remaining competitive in low-density regimes where semantic similarity suffices.

cs.CL↗

arXiv2026

Audio Spotforming via Post-Filtering Using Cross-Array Non-target Estimates

Audio spotforming is a technique for extracting target speech from noisy mixtures by utilizing multiple microphone arrays. Conventional methods estimate a shared target speech component from linearly separated signals obtained by each array using low-rank approximations and apply post filtering (PF) based on this estimated low-rank representation. However, owing to the mismatch between low-rank models and the complex structure of speech signals, directly relying on low-rank approximations for PF can degrade the speech extraction performance. In this study, we leverage the observation that non-target components located in the target speech direction from the perspective of one array can be spatially separated when viewed from other arrays. This insight motivates a new spotforming method for efficient post-filter estimation using non-target estimates across arrays instead of relying on low-rank approximations. Experiments demonstrate that the proposed method outperforms conventional spotforming methods.

cs.SD↗

arXiv2026

Privacy Protection Against Personalized Text-to-Image Synthesis via Cross-image Consistency Constraints

The rapid advancement of diffusion models and personalization techniques has made it possible to recreate individual portraits from just a few publicly available images. While such capabilities empower various creative applications, they also introduce serious privacy concerns, as adversaries can exploit them to generate highly realistic impersonations. To counter these threats, anti-personalization methods have been proposed, which add adversarial perturbations to published images to disrupt the training of personalization models. However, existing approaches largely overlook the intrinsic multi-image nature of personalization and instead adopt a naive strategy of applying perturbations independently, as commonly done in single-image settings. This neglects the opportunity to leverage inter-image relationships for stronger privacy protection. Therefore, we advocate for a group-level perspective on privacy protection against personalization. Specifically, we introduce Cross-image Anti-Personalization (CAP), a novel framework that enhances resistance to personalization by enforcing style consistency across perturbed images. Furthermore, we develop a dynamic ratio adjustment strategy that adaptively balances the impact of the consistency loss throughout the attack iterations. Extensive experiments on the classical CelebHQ and VGGFace2 benchmarks show that CAP substantially improves existing methods.

cs.CV↗

arXiv2026

Counterfactual Trace Auditing of LLM Agent Skills

Large Language Model agents are increasingly augmented with agent skills. Current evaluation methods for skills remain limited. Most deployed benchmarks report only pass rate before and after a skill is attached, treating the skill as a black box change to agent behavior. We introduce Counterfactual Trace Auditing (CTA), a framework for measuring how a skill changes agent behavior. CTA pairs each with skill agent trace with a without skill counterpart on the same task, segments both traces into goal directed phases, aligns the phases, and emits structured Skill Influence Pattern (SIP) annotations. These annotations describe the behavioral effect of a skill rather than only its task outcome. We instantiate CTA on SWE-Skills-Bench with Claude across 49 software engineering tasks. The resulting audit reveals a clear evaluation gap. Pass rate changes by only +0.3 percentage points on average, suggesting little aggregate effect. Yet CTA identifies 522 SIP instances across the same paired traces, showing that the skills substantially reshape agent behavior even when pass rate is nearly unchanged. The audit also separates several recurring effects that pass rate cannot detect, including literal template copying, off task artifact creation, excess planning, and task recovery. Three findings emerge. First, high baseline tasks contain most of the observed skill effects, although their pass rate is already saturated and therefore cannot reflect those effects. Second, tasks with moderate baseline performance show the most recoverable gain, but often at substantially higher token cost. Third, the dominant SIP type can be identified by baseline bucket: surface anchoring is most common on ceiling tasks and edge-case prompting is most common on mid-range and floor tasks. These regularities turn informal failure mode observations into reproducible behavioral measurements.

cs.AI↗