Know2Act: What Knowledge Reaches a Tool-Using Agent’s Actions?
Traces how predictive knowledge reaches tool-use decisions, with learned answer contracts for inference-time selection.
About
I am a Ph.D. student in Electrical Engineering at the University of Notre Dame. Previously, I received my B.S. in Computer Science from Beijing Jiaotong University.
I study self-improving and reliable LLM agents. My research asks when learning and experience actually improve an agent’s decisions: does acquired knowledge change its actions, and can useful experience transfer across tasks and models? I work on agentic post-training, experience memory, and reliable long-horizon behavior, using tool-using and interactive agents to examine the connection between learning, memory, and action.
My broader research spans LLM adaptation for human mobility, including Mobility-LLM (NeurIPS 2024), and deep learning for spatial sensing and environmental mapping.
Research
My long-term goal is to understand how agents turn experience into lasting, transferable improvements in reasoning and action. My current work connects Learning → Memory → Reliable Action.
Learning
I study how post-training and feedback change an agent’s decisions. Know2Act examines whether learned knowledge reaches tool-use policies; SARRA investigates how supervision affects task performance and resistance to prompt injection.
Memory
I examine when experience memory provides reusable knowledge across tasks and models. SkillRC audits memory gains to distinguish transferable experience from benefits tied to a particular executor and its context.
Reliable Action
I use controlled experiments and failure analysis to evaluate agent behavior under changing conditions. Building on this work, I aim to understand when lasting improvements require coordinated updates to the policy, memory, and execution structure.
Publications & manuscripts
Google Scholar ↗Current work on agents, alongside earlier work in spatial sensing and mobility.
SkillRC: Is It the Skill or the Stack? Auditing Experience Memory in Agent Harnesses
Separates memory content from context effects to understand how experience memory works across agent systems.
SARRA: Attack Resistance Is Only Half the Answer for Search Agents
Evaluates search-agent reliability through answer correctness, attack resistance, and explicit exposure and cost accounting.
Ultra-Wideband Indoor Localization and Bistatic Deep Learning-Based Mapping for ISAC
Under review at IEEE ICC 2027 · SAC – Integrated Sensing and Communications
Ultra-wideband localization and deep-learning-based environmental mapping for integrated sensing and communications.
Deep Learning Estimation of the Number and Locations of Reflective Surfaces for Underwater ISAC
Under review at IEEE ICASSP 2027
A multi-stage deep-learning and signal-processing pipeline for estimating reflective boundaries from noisy acoustic measurements.
Mobility-LLM: Learning Visiting Intentions and Travel Preferences from Human Mobility Data with Large Language Models
Adapts pretrained LLMs to human mobility through visiting-intention memory and travel-preference prompts.
Paper ↗LLM adaptation · Mobility
Teaching & Talks
75-MINUTE LECTURE
Retrieval-Augmented Generation
CSE 60556 LLM · University of Notre Dame
Where should knowledge live in an LLM system? From model parameters and context to external memory, this lecture introduces RAG, examines its limitations, and explores approaches including Self-RAG and GraphRAG.
SlidesTEACHING ASSISTANT
Introduction to Electrical and Computer Engineering
From Circuit Fundamentals to Physical Systems
Supported hands-on learning through lab instruction, student support, project debugging, and assessment.
Beyond Research
Meet Chocoliz (巧乐兹), my cat.
The familiar face hiding behind my portrait above.
