Artificial Intelligence · 2024 Cohort

Zhenhao Liu

Artificial Intelligence undergraduate at Westlake University working on automated scientific discovery and scientific agents.

2024 Cohort Hangzhou, China Continue reading
Aerial view of the Westlake University campus at dusk
Westlake University · Hangzhou, China Official website

Artificial Intelligence undergraduate at Westlake

I am an undergraduate student in Artificial Intelligence at Westlake University and conduct research at the Westlake NLP Lab under the guidance of Prof. Yue Zhang.

I study Artificial Intelligence as a member of the 2024 undergraduate cohort. My research focuses on automated scientific discovery with DeepScientist, scientific agents, and evidence-grounded evaluation of research systems.

Program
Artificial Intelligence
Cohort
2024
Focus
Automated scientific discovery, scientific agents, and research-system evaluation

Areas of interest

  1. Automated scientific discovery

    Using DeepScientist and scientific agents for reproducible workflows spanning problem formulation, experimentation, and evaluation.

  2. AI scientist evaluation

    Evidence-grounded evaluation of scientific-agent capabilities, research outputs, and human-AI responsibility boundaries.

  3. Research-system evaluation

    Evidence-grounded evaluation of AI-scientist capabilities, scientific originality, and human-AI responsibility.

Scientific evaluation and collaborative studies

Ministry of Education of China project · Data collection · System integration & evaluation

Evidence-Grounded Scientific Originality Evaluation System

Developed as part of a Ministry of Education of China project, this system evaluates the originality of scientific papers through document screening, parallel agent review, comparative evidence, and traceable final assessment.

Data
Collected and curated evaluation sets spanning papers from five universities, CNS and highly cited benchmarks, and expert-reviewed samples.
Integration and evaluation
Integrated the end-to-end evaluation pipeline and assessed the complete system through benchmark discrimination, bibliometric correlation, and expert-agreement analyses.

Research contributions

Additional collaborative publications

  • EMNLP 2026 Main Conference · Accepted · 2026

    How Far Are AI Scientists from Changing the World? A Survey of Autonomous Scientific Discovery

    Co-author · Data collection and curation OpenReview
  • EMNLP 2026 Main Conference · Accepted · 2026

    AI Drafts, Humans Judge: Quantifying the Human-AI Responsibility Boundary in Peer Review

    Co-author · Data collection and curation OpenReview

Task-specific representations for fMRI language encoding

Accepted poster, NLPCC 2026 Shared Tasks Sole first author

RepDuet at NLPCC 2026 Chinese BabyLM: Task-Specific Hybrid Representations for the Cognitive Track

Zhenhao Liu, Qiujie Xie, Minjun Zhu, Zhiyuan Ning, Yixuan Weng, Pinzhang Li, Jiajun Wang, and Yue Zhang

As sole first author, I designed task-specific hybrid representations for word- and sentence-level fMRI encoding, trained the models, and built controlled comparisons and stability analyses. I integrated DeepScientist into the end-to-end research workflow; after organizer reproduction, RepDuet ranked first in the Cognitive Track.

AI and programming competitions

I participate in AI and competitive programming competitions. These are selected results from the 2025 and 2026 seasons.

01
Grand Master Trophy
02
Gold medals
02
Silver medals
YearCompetition / SiteResultRecord

IOAI2 2026

AI Track

Grand Master Trophy -

ICPC

Asia East Continent Final

Silver Source

ICPC

Asia Xi'an Regional

Gold Source

CCPC

Zhengzhou Site

Gold Source

ICPC

Asia Shenyang Regional

Silver Source

Interested in a conversation?

I welcome thoughtful conversations around automated scientific discovery, scientific agents, and research-system evaluation.

liuzhenhao@westlake.edu.cn