LiveOak Compiler (2024Fall)
Chufeng Jiang
Implemented a compiler based on recursive-descent parsing written by Java
- A simple Java-Like language called LiveOak was compiled to a code for a stack machine named SaM and x86 assembly code
- Performed lexical analysis (scanning) and syntactic analysis (parsing) simultaneously to detect the grammar mistake and build an abstract syntax tree (AST)
- Traversed AST to perform semantic analysis includes type checking, formals and actuals checking, class & method checking, and register allocation for x86 assembly code.
- Performed code generation
Mixup-CLIPood: Robust Domain Generalization for Multi-modal Object Recognition (2024 AIEA)
Yuxin Qiao, Keqin Li, Junhong Lin, Rong Wei, Chufeng Jiang, Yang Luo, Haoyu Yang.
Contribution:
- Address the incongruity between the actual loss and the one documented, and we deduce the actual loss used.
- Expand the experiments to encompass two larger vision-language backbones.
- Propose “Mixup-CLIPood” with a novel mix-up loss to enhance the previous model’s generalization ability.
Large language models for forecasting and anomaly detection: A systematic literature review (2024 Arxiv)
Jing Su, Chufeng Jiang, Xin Jin, Yuxin Qiao, Tingsong Xiao, Hongda Ma, Rong Wei, Zhi Jing, Jiajun Xu, Junhong Lin
Research Questions:
- Q1: What methodologies are employed in LLMs for forecasting in different domains?
- Q2: How effective are LLMs in detecting anomalies compared to traditional anomaly detection methods?
- Q3：What are the limitations and challenges of using LLMs for forecasting and anomaly detection?.
Academic Impact:
