CV
Yanzhou Mu
Doctoral Student in Software Engineering
Summary
Doctoral student in Software Engineering at Nanjing University. Research direction: software testing, with work on deep learning framework testing, model mutation, metamorphic testing, heuristic scheduling, and testing tools for MindSpore and PyTorch-based workflows.
Education
- Ph.D. Student in Software Engineering2026.06Nanjing University, Software College
- Master in Software Engineering2022.01Tianjin University, Department of Intelligence and Computing
- Bachelor in Computer Science and Technology2019.06Nantong University, School of Computer Science and Technology
Skills
Research Areas
- Software testing
- Deep learning framework testing
- Model mutation
- Metamorphic testing
- Heuristic scheduling
- MindSpore
- PyTorch
Publications
- DevMuT: Testing Deep Learning Framework via Developer Expertise-Based Mutation2024IEEE/ACM Automated Software Engineering Conference (ASE 2024)CCF-A conference publication.
- Improving Deep Learning Framework Testing with Model-Level Metamorphic Testing2025International Symposium on Software Testing and Analysis (ISSTA 2025)CCF-A conference publication.
- Deep Learning Framework Testing via Model Mutation: How far are we?2025
- Understanding LLM-Centric Challenges for Deep Learning Frameworks: An Empirical Analysis2026
Portfolio
- MindSpore Network Generalization Testing TechnologyProjectCompleted project. Student leader. Developed a MindSpore-based network structure generalization testing tool.
- Research on Key Mutation Testing Techniques for DL FrameworksProjectOngoing project. Student leader. Designed mutation rules and heuristic guidance for deep learning framework testing.
- Fast Detection, Localization, and Patch Generation for Concurrency DefectsProjectCompleted project. Project participant. Designed a heuristic scheduling testing method for Java concurrent programs.
- LLM Generalization Testing Tool for MindSporeProjectOngoing project. Student leader. Developed a component-based generalization testing tool for large-scale models on MindSpore.