Searcharxiv⌕ Search

arXiv subjects

Li Chen

Publications and source records attributed to Li Chen.

At least 91 records · Page 5Linked to original sources
arXiv2026

Human-Less LLM Serving: Quantifying the Human Tax on Throughput

Every major LLM serving system is designed to meet TTFT and TPOT SLOs. These metrics capture latency as a human user perceives it, and the mechanisms built to satisfy them are now standard infrastructure. We observe that long-horizon AI tasks call LLMs programmatically in tight loops where no human observes TTFT or TPOT. We ask: how much throughput do serving systems sacrifice to meet TTFT and TPOT SLAs that these workloads never need? We conduct a systematic measurement study across chunk sizes, SLO settings, context lengths, and concurrency levels. We find that the human tax on throughput grows substantially with context length and lands in the 60-93% range. At 64K token contexts, tightening the TTFT SLO to production-typical settings costs a large fraction of throughput versus the human-less baseline. The human tax is larger at higher concurrency and is qualitatively similar across SGLang and Sarathi-Serve. We term the unconstrained optimum human-less serving and provide a prototype demonstrating that it is practical on real workloads. Our findings argue that serving systems should expose workload-class-aware SLA configurations rather than silently applying the human tax uniformly to all traffic.

cs.NI↗
arXiv2026

ReSim: Reliable World Simulation for Autonomous Driving

How can we reliably simulate future driving scenarios under a wide range of ego driving behaviors? Recent driving world models, developed exclusively on real-world driving data composed mainly of safe expert trajectories, struggle to follow hazardous or non-expert behaviors, which are rare in such data. This limitation restricts their applicability to tasks such as policy evaluation. In this work, we address this challenge by enriching real-world human demonstrations with diverse non-expert data collected from a driving simulator (e.g., CARLA), and building a controllable world model trained on this heterogeneous corpus. Starting with a video generator featuring a diffusion transformer architecture, we devise several strategies to effectively integrate conditioning signals and improve prediction controllability and fidelity. The resulting model, ReSim, enables Reliable Simulation of diverse open-world driving scenarios under various actions, including hazardous non-expert ones. To close the gap between high-fidelity simulation and applications that require reward signals to judge different actions, we introduce a Video2Reward module that estimates a reward from ReSim's simulated future. Our ReSim paradigm achieves up to 44% higher visual fidelity, improves controllability for both expert and non-expert actions by over 50%, and boosts planning and policy selection performance on NAVSIM by 2% and 25%, respectively.

cs.CV↗
arXiv2026

MemRec: Collaborative Memory-Augmented Agentic Recommender System

The evolution of recommender systems has shifted from traditional collaborative filtering to LLM-based agentic systems, which rely on semantic user and item memories to make predictions. However, existing agents maintain these memories in isolation. This overlooks crucial collaborative signals, such as user-item co-engagements and peer relationships across the community, which significantly limits their ability to uncover hidden preferences and accurately infer user needs, particularly for data-sparse users. To bridge this gap, we introduce collaborative memory, a paradigm that connects isolated semantics to enable the sharing of relational insights. Yet, naively utilizing collaborative memory causes severe context overload and introduces noise to downstream LLMs, alongside prohibitive computational costs. To resolve this, we propose MemRec, a framework that architecturally decouples memory management from reasoning. MemRec introduces a dedicated, lightweight language model (LM_Mem) to efficiently manage and synthesize a dynamic collaborative memory graph in the background. It provides only distilled, high-signal contexts to a downstream, heavyweight large language model (LLM_Rec) for the final recommendation. Extensive experiments on four benchmarks demonstrate that MemRec achieves state-of-the-art performance. Code: https://github.com/rutgerswiselab/memrec and Homepage: https://memrec.weixinchen.com/

cs.IR↗
arXiv2026

RISE: Self-Improving Robot Policy with Compositional World Model

Despite the sustained scaling on model capacity and data acquisition, Vision-Language-Action (VLA) models remain brittle in contact-rich and dynamic manipulation tasks, where minor execution deviations can compound into failures. While reinforcement learning (RL) offers a principled path to robustness, on-policy RL in the physical world is constrained by safety risk, hardware cost, and environment reset. To bridge this gap, we present RISE, a scalable framework of robotic reinforcement learning via imagination. At its core is a Compositional World Model that (i) predicts multi-view future via a controllable dynamics model, and (ii) evaluates imagined outcomes with a progress value model, producing informative advantages for the policy improvement. Such compositional design allows state and value to be tailored by best-suited yet distinct architectures and objectives. These components are integrated into a closed-loop self-improving pipeline that continuously generates imaginary rollouts, estimates advantages, and updates the policy in imaginary space without costly physical interaction. Across three challenging real-world tasks, RISE yields significant improvement over prior art, with more than +35% absolute performance increase in dynamic brick sorting, +45% for backpack packing, and +35% for box closing, respectively.

cs.RO↗
arXiv2026

Finite-Precision Conjugate Gradient Method for Massive MIMO Detection

The implementation of the conjugate gradient (CG) method for massive MIMO detection is computationally challenging, especially for a large number of users and correlated channels. In this paper, we propose a low computational complexity CG detection from a finite-precision perspective. First, we develop a finite-precision CG (FP-CG) detection to mitigate the computational bottleneck of each CG iteration and provide the attainable accuracy, convergence, and computational complexity analysis to reveal the impact of finite-precision arithmetic. A practical heuristic is presented to select suitable precisions. Then, to further reduce the number of iterations, we propose a joint finite-precision and block-Jacobi preconditioned CG (FP-BJ-CG) detection. The corresponding performance analysis is also provided. Finally, simulation results validate the theoretical insights and demonstrate the superiority of the proposed detection.

eess.SP↗
arXiv2026

Causality-Encoded Diffusion Models for Interventional Sampling and Edge Inference

Standard diffusion models are flexible estimators of complex distributions, but they do not encode causal structures and therefore do not by themselves support causal analysis. We propose a causality-encoded diffusion framework that incorporates a known directed acyclic graph by training conditional diffusion models consistent with the graph factorisation. The resulting sampler approximately recovers the observational distribution and enables interventional sampling by fixing intervened variables while propagating effects through the graph during reverse diffusion. Building on this interventional simulator, we develop a resampling-based test for directed edges that generates null replicates under a candidate graph. We establish convergence guarantees for observational and interventional distribution estimation, with rates governed by the maximum local dimension rather than the ambient dimension, and prove asymptotic control of type I error for the edge test. Simulations show improved interventional distribution recovery relative to baselines, with near-nominal size and favourable power in inference. An application to flow cytometry data demonstrates practical utility of the proposed method in assessing disputed signalling linkages.

stat.ME↗
arXiv2026

Designing Heaven's Will: The job assignment in the Chinese imperial civil service

We provide an original analysis of historical documents to describe the assignment procedures used to allocate entry-level civil service jobs in China from the tenth to the early twentieth century. The procedures tried to take different objectives into account through trial and error. By constructing a formal model that combines these procedures into a common framework, we compare their effectiveness in minimizing unfilled jobs and prioritizing high-level posts. We show that the problem was inherently complex such that changes made to improve the outcome could have the opposite effect. Based on a small modification of the last procedure used, we provide a new mechanism for producing maximum matchings under constraints in a transparent and public way.

econ.TH↗
arXiv2026

SAGER: Self-Evolving User Policy Skills for Recommendation Agent

Large language model (LLM) based recommendation agents personalize what they know through evolving per-user semantic memory, yet how they reason remains a universal, static system prompt shared identically across all users. This asymmetry is a fundamental bottleneck: when a recommendation fails, the agent updates its memory of user preferences but never interrogates the decision logic that produced the failure, leaving its reasoning process structurally unchanged regardless of how many mistakes it accumulates. To address this bottleneck, we propose SAGER (Self-Evolving Agent for Personalized Recommendation), the first recommendation agent framework in which each user is equipped with a dedicated policy skill, a structured natural-language document encoding personalized decision principles that evolves continuously through interaction. SAGER introduces a two-representation skill architecture that decouples a rich evolution substrate from a minimal inference-time injection, an incremental contrastive chain-of-thought engine that diagnoses reasoning flaws by contrasting accepted against unchosen items while preserving accumulated priors, and skill-augmented listwise reasoning that creates fine-grained decision boundaries where the evolved skill provides genuine discriminative value. Experiments on four public benchmarks demonstrate that SAGER achieves state-of-the-art performance, with gains orthogonal to memory accumulation, confirming that personalizing the reasoning process itself is a qualitatively distinct source of recommendation improvement.

cs.IR↗
arXiv2026

Rethinking AI Hardware: A Three-Layer Cognitive Architecture for Autonomous Agents

The next generation of autonomous AI systems will be constrained not only by model capability, but by how intelligence is structured across heterogeneous hardware. Current paradigms -- cloud-centric AI, on-device inference, and edge-cloud pipelines -- treat planning, reasoning, and execution as a monolithic process, leading to unnecessary latency, energy consumption, and fragmented behavioral continuity. We introduce the Tri-Spirit Architecture, a three-layer cognitive framework that decomposes intelligence into planning (Super Layer), reasoning (Agent Layer), and execution (Reflex Layer), each mapped to distinct compute substrates and coordinated via an asynchronous message bus. We formalize the system with a parameterized routing policy, a habit-compilation mechanism that promotes repeated reasoning paths into zero-inference execution policies, a convergent memory model, and explicit safety constraints. We evaluate the architecture in a reproducible simulation of 2000 synthetic tasks against cloud-centric and edge-only baselines. Tri-Spirit reduces mean task latency by 75.6 percent and energy consumption by 71.1 percent, while decreasing LLM invocations by 30 percent and enabling 77.6 percent offline task completion. These results suggest that cognitive decomposition, rather than model scaling alone, is a primary driver of system-level efficiency in AI hardware.

cs.AI↗
arXiv2026

M3R: Localized Rainfall Nowcasting with Meteorology-Informed MultiModal Attention

Accurate and timely rainfall nowcasting is crucial for disaster mitigation and water resource management. Despite recent advances in deep learning, precipitation prediction remains challenging due to limitations in effectively leveraging diverse multimedia data sources. We introduce M3R, a Meteorology-informed MultiModal attention-based architecture for direct Rainfall prediction that synergistically combines visual NEXRAD radar imagery with numerical Personal Weather Station (PWS) measurements, using a comprehensive pipeline for temporal alignment of heterogeneous meteorological data. With specialized multimodal attention mechanisms, M3R novelly leverages weather station time series as queries to selectively attend to spatial radar features, enabling focused extraction of precipitation signatures. Experimental results for three spatial areas of 100 km * 100 km centered at NEXRAD radar stations demonstrate that M3R outperforms existing approaches, achieving substantial improvements in accuracy, efficiency, and precipitation detection capabilities. Our work establishes new benchmarks for multimedia-based precipitation nowcasting and provides practical tools for operational weather prediction systems. The source code is available at https://github.com/Sanjeev97/M3Rain

cs.LG↗
arXiv2026

Relaxation dynamics of the continuum Kuramoto model with non-integrable kernels

We study the asymptotic behavior of the continuum Kuramoto model with a fractional Laplacian-type kernel. For this, we construct global weak solutions via a two-parameter regularization procedure using a kernel truncation with fractional dissipation. Using a priori uniform estimates derived in fractional Sobolev spaces, we employ compactness arguments to construct global weak solutions to the singular continuum Kuramoto model. Furthermore, we also establish an exponential relaxation toward the initial phase average in $L^2$-norm under suitable assumptions on initial data and system parameters. These findings provide a rigorous characterization of the existence of solutions and the emergent dynamics of Kuramoto ensembles under physically important strongly singular interactions, including power-law singular kernels and Coulomb-type kernels.

math.AP↗
arXiv2026

An Adaptive Antenna Impedance Matching Method via Deep Reinforcement Learning

Adaptive impedance matching between antennas and radio frequency front-end modules is critical for maximizing power transmission efficiency in mobile communication systems. Conventional numerical and analytical methods struggle with a trade-off between accuracy and efficiency, while deep neural network (DNN)-based supervised learning approaches rely heavily on large labeled datasets and lack flexibility for dynamic environments. To address these limitations, this paper proposes a deep reinforcement learning (DRL)-based approach for adaptive impedance matching. First, we model the impedance tuning problem as an optimal control problem, proving the feasibility of solving the optimal control law via reinforcement learning. Then, we design a tailored DRL framework for impedance tuning, which employs a compact state representation that integrates key frequency characteristics and matching quality metrics. Additionally, this framework incorporates a piecewise reward function that accounts for both matching accuracy and tuning speed. Furthermore, a test-phase exploration mechanism is introduced to enhance tuning stability, which effectively reduces local optimal trapping and high-frequency tuning variance. Experimental results demonstrate that the proposed method achieves superior performance in terms of tuning accuracy, efficiency, and stability compared with conventional heuristic and gradient-based methods, making it promising for practical impedance tuning systems.

eess.SP↗
arXiv2026

TAMEn: Tactile-Aware Manipulation Engine for Closed-Loop Data Collection in Contact-Rich Tasks

Handheld paradigms offer an efficient and intuitive way for collecting large-scale demonstration of robot manipulation. However, achieving contact-rich bimanual manipulation through these methods remains a pivotal challenge, which is substantially hindered by hardware adaptability and data efficacy. Prior hardware designs remain gripper-specific and often face a trade-off between tracking precision and portability. Furthermore, the lack of online feasibility checking during demonstration leads to poor replayability. More importantly, existing handheld setups struggle to collect interactive recovery data during robot execution, lacking the authentic tactile information necessary for robust policy refinement. To bridge these gaps, we present TAMEn, a tactile-aware manipulation engine for closed-loop data collection in contact-rich tasks. Our system features a cross-morphology wearable interface that enables rapid adaptation across heterogeneous grippers. To balance data quality and environmental diversity, we implement a dual-modal acquisition pipeline: a precision mode leveraging motion capture for high-fidelity demonstrations, and a portable mode utilizing VR-based tracking for in-the-wild acquisition and tactile-visualized recovery teleoperation. Building on this hardware, we unify large-scale tactile pretraining, task-specific bimanual demonstrations, and human-in-the-loop recovery data into a pyramid-structured data regime, enabling closed-loop policy refinement. Experiments show that our feasibility-aware pipeline significantly improves demonstration replayability, and that the proposed visuo-tactile learning framework increases task success rates from 34% to 75% across diverse bimanual manipulation tasks. We further open-source the hardware and dataset to facilitate reproducibility and support research in visuo-tactile manipulation.

cs.RO↗
arXiv2026

TR-ICRL: Test-Time Rethinking for In-Context Reinforcement Learning

In-Context Reinforcement Learning (ICRL) enables Large Language Models (LLMs) to learn online from external rewards directly within the context window. However, a central challenge in ICRL is reward estimation, as models typically lack access to ground-truths during inference. To address this limitation, we propose Test-Time Rethinking for In-Context Reinforcement Learning (TR-ICRL), a novel ICRL framework designed for both reasoning and knowledge-intensive tasks. TR-ICRL operates by first retrieving the most relevant instances from an unlabeled evaluation set for a given query. During each ICRL iteration, LLM generates a set of candidate answers for every retrieved instance. Next, a pseudo-label is derived from this set through majority voting. This label then serves as a proxy to give reward messages and generate formative feedbacks, guiding LLM through iterative refinement. In the end, this synthesized contextual information is integrated with the original query to form a comprehensive prompt, with the answer determining through a final round of majority voting. TR-ICRL is evaluated on mainstream reasoning and knowledge-intensive tasks, where it demonstrates significant performance gains. Remarkably, TR-ICRL improves Qwen2.5-7B by 21.23% on average on MedQA and even 137.59% on AIME2024. Extensive ablation studies and analyses further validate the effectiveness and robustness of our approach. Our code is available at https://github.com/pangpang-xuan/TR_ICRL.

cs.CL↗
arXiv2026

SMASH: Mastering Scalable Whole-Body Skills for Humanoid Ping-Pong with Egocentric Vision

Existing humanoid table tennis systems remain limited by their reliance on external sensing and their inability to achieve agile whole-body coordination for precise task execution. These limitations stem from two core challenges: achieving low-latency and robust onboard egocentric perception under fast robot motion, and obtaining sufficiently diverse task-aligned strike motions for learning precise yet natural whole-body behaviors. In this work, we present \methodname, a modular system for agile humanoid table tennis that unifies scalable whole-body skill learning with onboard egocentric perception, eliminating the need for external cameras during deployment. Our work advances prior humanoid table-tennis systems in three key aspects. First, we achieve agile and precise ball interaction with tightly coordinated whole-body control, rather than relying on decoupled upper- and lower-body behaviors. This enables the system to exhibit diverse strike motions, including explosive whole-body smashes and low crouching shots. Second, by augmenting and diversifying strike motions with a generative model, our framework benefits from scalable motion priors and produces natural, robust striking behaviors across a wide workspace. Third, to the best of our knowledge, we demonstrate the first humanoid table-tennis system capable of consecutive strikes using onboard sensing alone, despite the challenges of low-latency perception, ego-motion-induced instability, and limited field of view. Extensive real-world experiments demonstrate stable and precise ball exchanges under high-speed conditions, validating scalable, perception-driven whole-body skill learning for dynamic humanoid interaction tasks.

cs.RO↗
arXiv2026

Three Years with Classroom AI in Introductory Programming: Shifts in Student Awareness, Interaction, and Performance

Generative AI (GenAI) tools such as ChatGPT now provide novice programmers with instant, personalized support and are reshaping computing education. While a growing body of work examines AI's immediate impacts, longitudinal evidence remains limited on how students' awareness, student-AI interaction patterns, and course outcomes evolve as AI becomes routine in classrooms. To address this gap, we investigate an introductory Python course across three successive AI-supported cohorts (2023-2025). Using questionnaires, coded student-AI dialogue logs, and course assessment records, we examine cohort-to-cohort shifts in students' AI awareness, interaction practices, and learning outcomes. We find that students' relationships with GenAI change systematically over time: familiarity and uptake become increasingly normative, and help-seeking practices evolve alongside growing AI literacy and shifting expectations of what the assistant should provide. These changes suggest that, in the AI era, the central instructional challenge is less about whether students use AI and more about how courses redefine productive learning practices while maintaining student agency. Our study offers longitudinal evidence and practical implications for designing and integrating AI programming support in course settings.

cs.HC↗
arXiv2026

Design Implications for Student and Educator Needs in AI-Supported Programming Learning Tools

AI-powered coding assistants can support students in programming courses by providing on-demand explanations and debugging help. However, existing research often focuses on individual tools, leaving a gap in evidence-based design recommendations that reflect both educator and student perspectives in education settings. To ground the design of learning-oriented AI coding assistants for both sides' needs, we conducted parallel surveys of educators (N=50) and students (N=90) to compare preferences about (i) how students should request help, (ii) how AI should respond, and (iii) who should control. Our results show that educators generally favored indirect scaffolding that preserves students' reasoning, whereas students were more likely to prefer direct, actionable help. Educators further highlighted the need for course-aligned constraints and instructor-facing oversight, while students emphasized timely support and clarity when stuck. Based on these findings, we discuss the interaction-focused design space and derive design implications for learning-oriented AI coding assistants, highlighting scaffolding and control mechanisms that balance students' agency with instructional constraints.

cs.HC↗
arXiv2026

Designing a Meta-Reflective Dashboard for Instructor Insight into Student-AI Interactions

Generative AI tools are increasingly used for coursework help, shifting much of students' help-seeking and reasoning into student-AI chats that are largely invisible to instructors. This loss of visibility can weaken instructors' ability to understand students' difficulties, ensure alignment with course goals, and uphold course policies. Yet transcript-level access is neither scalable nor ethically straightforward: reading raw chat logs across a class is impractical, and exposing detailed dialogue can raise privacy concerns and chilling effects on help seeking. As a result, instructors face a tension between needing actionable insight and avoiding default surveillance of student conversations. To address this gap, we propose a meta-reflective dashboard that makes student-AI sessions interpretable without exposing raw chat logs by default. After each help-seeking session, a reflection AI produces a structured, session-level summary of the student's interaction trajectory, AI usage patterns, and potential risks. We co-designed the dashboard with instructors and students to surface key challenges and design goals, and conducted a formative evaluation of perceived usefulness, trust in the summaries, and privacy acceptability. Findings suggest that the proposed dashboard can reduce instructors' sensemaking effort while mitigating privacy concerns associated with transcript-level access, and they also yield design implications for evidence, governance, and scalable class-level analytics for AI-supported learning.

cs.HC↗