👋 Hi, I'm Junqi Mao!
🚀 Passionate about AI, competitive programming, and computational medicine.
🎓 Bachelor's in Statistics (Data Science & AI), Shandong University.
🏆 ICPC Regional Silver Medalist
🔍 Interested in interpretable AI for medical applications.
💡 Experienced in deep learning (PyTorch, TensorFlow), algorithm optimization, and software development.
📂 Check out my projects:
Project Overview:
This project delves into the network maximum flow problem, a pivotal combinatorial optimization challenge in graph theory. It begins by establishing the fundamental symbols and definitions of the maximum flow problem and then categorizes existing methods into two main approaches: augmenting path algorithms and preflow-push algorithms.
For the augmenting path approaches, the project introduces the EK/Dinic algorithms and their optimizations, along with the MPM/ISAP variants and a comparative performance analysis. For preflow-push approaches, it focuses on analyzing the characteristics of the HLPP algorithm and proposes an improved Budget Algorithm to mitigate the "ping-pong effect", validated through extensive comparative experiments.
Additionally, the project summarizes four common problem reduction paradigms:
- Bipartite Graph Maximum Matching
- Minimum-Cost Maximum Flow
- Bipartite Graph Maximum Weight Matching
- Minimum Cut Problem
All implementations are done in C++ with comprehensive testing, making the project a valuable resource for both theoretical research and practical applications.
Tech Stack:
- C++
- Algorithm Optimization & Performance Testing
Project Overview: This project focuses on testing the performance of various segmentation algorithms using PaddleSeg, an end-to-end image segmentation toolkit based on PaddlePaddle. It is designed to evaluate multiple segmentation models across key metrics such as accuracy, efficiency, and robustness in medical imaging tasks. By benchmarking a diverse range of pre-trained models and custom configurations, the project aims to identify optimal strategies for medical image segmentation, facilitating improvements in both research and practical applications.
Tech Stack:
- python
- PaddlePaddle & PaddleSeg