Biography
Hi ! This is Qianqi Yan (严乾琪), a senior undergraduate student majoring in Computer Science and minoring in Mathematics. I expect to receive my bachelor’s degrees from University of Michigan College of Engineering and Shanghai Jiao Tong University - University of Michigan Joint Institute.
Currently, I’m working at Situated Language and Emboddied Dialogue (SLED) lab advised by Professor Joyce Chai and Jianing “Jed” Yang. I’m also advised by Professor Stella Yu and Tsung-Wei Ke.
- Selected Courses: Natural Language Processing (A), Deep Learning for Computer Vision (A+), Machine Learning (A), Computer Organization (A), Data Structure & Algorithm (A)
Download my resumé.
- Natural Language Processing
- Deep Learning
- Computer Vision
B.S.E. in Computer Science, 2023(expected)
University of Michigan, Ann Arbor
B.S.E. in Electrical and Computer Engineering, 2023(expected)
Shanghai Jiao Tong University
Experience
Advised by Stella Yu
Responsibilities include:
- Create a vision model which conducts image-level recognition and segmentation concurrently.
- Construct the hierarchy of image segmentation and language entity separately, and then match them at each level of granularity to derive hierarchical semantic segmentation.
Advised by Joyce Chai
Responsibilities include:
- Leverage language models to resolve uncertainty in vision (disambiguating).
Advised by Joyce Chai and Jianing “Jed” Yang
Responsibilities include:
- Design the prompting pipeline to query GPT-3 to generate actionable plans based on the goal state and environment feedback.
- Collect 65k egocentric view - text description of goal state pairs from the FILM dataset and fine-tune CLIP model to match goal states with stored frames during inference.
Advised by Rada Mihalcea and Spencer Vagg
Responsibilities include:
- Suggesting and implementing methods to address people references and improve quality of summarization output
- Deploying pipeline to transcribe recorded sales conversations and summarize them
- Design a service robot to be applied in hospital wards which is capable of navigating itself in unknown environment based on SLAM algorithm and providing customized message to patients according to their locations in the ward.