Huang
50 indexed papers
Publications per year
Top categories
Frequent co-authors
Research Timeline
This paper introduces Continual Learning Exam (CLE), a large-scale benchmark to measure experience-driven improvement in LLM agents.
Proposes a three-stage analytical framework for SoC floorplanning that enforces constraints more rigorously and achieves higher-quality solutions.
The paper studies sampling and recovery for tensor-valued dynamical systems using partial spatial observations over time.
This paper introduces Sluice, an adaptive data stream processing system for hybrid clouds, compatible with dataflow streaming engines.
Proposes a semantic successive cancellation list decoder for polar codes that incorporates a fine-tuned language model to improve communication reliability.
This paper investigates whether multi-agent large language models can exhibit emergent behaviors through interaction and introduces a framework to quantify this effect.
This paper introduces ThinkWorld, a VLA learning framework that grounds action reasoning in world dynamics and strengthens it through reinforcement learning with world imagination.
This paper proposes a dynamic scheduling framework, adaptive morsel scheduling (AMS), to improve the performance of sparse general matrix-matrix multiplication on skewed workloads.
The paper introduces InfraDev, a benchmark for evaluating AI infrastructure development, and shows that existing benchmarks are inadequate for assessing real-world tasks.
This paper proposes GradCO, a method for text-based person anomaly search that preserves identity cues and distinguishes fine-grained behaviors.
The paper proposes TaskBridge, a framework for reformulating task-oriented queries into product retrieval expressions.
The paper proposes a new framework for channel estimation in fluid antenna systems using a U-shaped Fourier neural operator.
This paper investigates the effectiveness of explicit reasoning in large language models for code generation and bug fixing.
One clear sentence describing what this paper does
This paper proposes RadioTraj3D, a method for constructing three-dimensional radio maps using building-height information to guide measurement selection and reconstruction.
RFLocMatch improves radio fingerprint localization by reducing spatial ambiguity and measurement perturbations.
This paper introduces BEAKER, a benchmark for evaluating embodied AI capabilities in self-driving chemical laboratories.
The paper introduces FrontierPhysics, an end-to-end benchmark for evaluating AI agents in physics research.
The paper proposes PSGAgent, a framework that enhances visual perception for Panoptic Scene Graph generation through a think-act-observe loop.
WeLM-Image is a unified image foundation model for text-to-image generation, instruction-guided editing, and multi-reference synthesis.
Papers
PSGAgent: Enhancing MLLMs' Dynamic Visual Perception for Panoptic Scene Graph Generation via Agentic Reasoning
Qingyun Liu, Yuankun Liu, Jiaye Feng, Jie Huang +5 more
The paper proposes PSGAgent, a framework that enhances visual perception for Panoptic Scene Graph generation through a think-act-observe loop.