Lei
50 indexed papers
Research Timeline
This paper introduces CW-Ghost, a method for estimating cache line fill volume and determining helper-thread prefetching granularity based on cache capacity constraints.
Argus is a persistent, self-evolving runtime system that achieves better long-horizon reasoning results than Direct Copilot in seven GPT-5.5 benchmark arenas, with improvements in solve-input tokens, active workflow time, and verification-gated self-evolution.
This paper introduces TrajDebug, an error-lifecycle tracing framework for LLM-based agentic systems, which addresses long-trajectory error discovery using multi-granularity history compression and evidence-based error identification.
This paper proposes SemBridge, a framework for continuous-latent autoregressive speech generation using discrete semantic tokens for supervision.
This paper reviews the NTIRE 2026 Low-light Enhancement competition, which merged misaligned smartphone images into a single, clean image in low-light conditions, achieving improvements of up to +6.49 dB in PSNR and +0.0101 in SSIM.
This paper introduces OpenCodeReview, a deterministic code review system that addresses non-determinism and context locality issues in LLM-based agents, achieving higher SEM-F1 scores and consuming fewer tokens.
This paper introduces GameLogicBench, a benchmark for evaluating the ability of coding agents to implement gameplay rules in game development.
The paper proposes a method called ForgetMimic to eliminate specific motions from learned policies in humanoid control while preserving the effectiveness of other motions.
The paper summarizes the Interspeech2026 Multilingual Conversational Speech Language Model Challenge and presents the results and insights from the challenge.
This paper introduces Grounded Entity Biographies (GEB), a long-video memory framework that groups visually grounded observations of the same physical instance across clips into retrievable biographies.
This paper proposes a pretraining framework called NeuroDyn-EEG that integrates generative priors from neural dynamics to estimate anatomically indexed physiological parameters from scalp EEG.
This paper improves the efficiency of distributed quantum computing by concurrently executing multiple compilations of the same quantum circuit.
The paper introduces MTEdit, a multi-turn image editing dataset and benchmark, to improve models' multi-turn image editing capabilities.
The paper presents a system called SkyAgent equipped with Revisable Target Memory (RTM) that improves continuous aerial search performance by revising and updating target hypotheses.
A framework for measuring human contributions in human-agent joint production is proposed.
This paper introduces RealComm, a dataset and augmentation pipeline for evaluating the effects of real call transmission on audio deepfake detection.
This paper introduces Sluice, an adaptive data stream processing system for hybrid clouds, compatible with dataflow streaming engines.
This paper evaluates the spatial capabilities of visual foundation models through four levels of spatial capability and proposes a new evaluation framework.
The authors prove that the unweighted decision problem of partial interval multicover is NP-complete even for unit intervals.
The paper introduces Grounded Generative Speech Recognition (GSR), a reformulation of speech recognition as generating semantically grounded text instead of a single transcript.
Papers
Partial Interval Multicover Is NP-Complete Even for Unit Intervals
Mingchao Zhou, Yingli Ran, Lei Zhao, Xiangzhi Tu +1 more
The authors prove that the unweighted decision problem of partial interval multicover is NP-complete even for unit intervals.