Scholar
Ziyan Luo
Google Scholar ID: C77ZMLAAAAAJ
PhD Candidate, Mila, McGill
Citations & Impact
Citations
45
H-index
3
i10-index
1
Publications
8
Co-authors
12
list available
Publications
Resume (English only)
Academic Achievements
- Paper 'Understanding Behavioral Metric Learning': large-scale study unifying and evaluating behavioral metric learning across 370 RL tasks under diverse noise settings
- Survey 'Discovering Temporal Structure: An Overview of Hierarchical Reinforcement Learning': 80+ page comprehensive overview focusing on temporal abstraction methods
- ACN paper accepted by WSDM'20 (top 7%, oral presentation)
- RLMob paper accepted by WSDM'22
- Reviewer for CAV'24, NeurIPS'24, TMLR, and NeurIPS'23 MathAI workshop
- Volunteer at CAV'24
- Teaching Assistant for COMP 579 (Reinforcement Learning) at McGill; recipient of McGill TA Award (awarded to 4 TAs annually department-wide)
Background
- Ph.D. candidate at Mila, McGill University, advised by Dr. Xujie Si and Dr. Doina Precup
- Broadly interested in Reinforcement Learning (RL), with a current focus on representation learning—developing compact, structured, and agent-centric encodings to support generalization, sample-efficient learning, and planning
- Central research theme: understanding how abstraction can serve as a foundation for representation learning, drawing inspiration from formal verification’s use of equivalence and refinement to manage complexity
- Explores behavioral metrics for state abstraction and temporal abstractions in hierarchical RL to integrate formal structure with learning-based flexibility
- Inspired by the formal methods community’s emphasis on precision, scientific rigor, and long-term research impact
