Updated on October 23, 2022
CV (pdf)
About
I am a PhD student in the Computer Science & Engineering at the University of Michigan, working with
Prof. Jenna Wiens.
My research lies at the intersection of causality and healthcare.
- PhD in Computer Science & Engineering
- Sepetember 2023 - June 2028 (Expected)
- B.S in Electrical Engineering, Cum Laude
- Sepetember 2018 - June 2022
- GPA: 3.8/4.0
- Winston Chen, William Stafford Noble, Yang Young Lu, "DeepROCK: Error-controlled interaction detection in deep neural networks", arxiv. 2023.
Link.
Poster.
Research Experience
Confidence Estimation for Network Propagation
- Developed a framework to estimate statistical confidence (q-value) for a class of ML algorithm called network propagation.
- Implemented the framework on Rankprop, a network propagation-based protein homology detection algorithm.
- Applied Rankprop with confidence estimation on a large-scale protein database (SCOP) to demonstrate its ability of identifying protein homologies with high-precision.
- Presented research work at University of Washington's annual undergraduate research symposium.
Error-controlled Interaction Detection in Neural Network
- Developed a neural network interpretation method that discovers learned feature interactions at desired error rate.
- Designed and implemented a compute module that debiases the raw interaction interpretation results.
- Applied the interpretation method on fruit flies (Drosophila) genomics data to identify meaningful transcription factor (TF) interactions.
Real-Time SDR to HDR Up Conversion
- Conducted research on a class of SDR to HDR video up-conversion algorithm called inverse tone mapping.
- Improved the existing inverse tone mapping algorithm by introducing a dynamic parameter estimation based on input image.
- Built a webpage for interactively optimizing inverse tone mapping parameter for given images.
- Summarized the research result in a poster and presented at UW's 2022 ENGINE show case.
Honors and Awards
Teaching Experience
Industry Experience
Leadership Experience
Technical Skills
- Programming Languages
-
Python
, JavaScript
, Java
, C/C++
, R
- Frameworks
-
PyTorch
, TensorFlow
, Numpy
, NetworkX
- Tools
Projects
Posters & Presentations
2022 ENGINE Showcase
- Motivation of this work: developed a real-time SDR to HDR conversion algorithm that can be applied to perform video conversion on the fly.
- Discussed the two existing approach to perform SDR to HDR and their limitations: 1) Inverse tone mapping(ITM)-based approach (low conversion quality). 2) Deep learning-based approach (too slow for real-time application)
- Explained our approach: an improved ITM method that can dynamically estimate the optimal conversion parameters using ML techniques.
NVIDIA Internal Meeting
A Log-based Testing Tool for NVIDIA's Genomics Computing Software, Sept. 10, 2021
- Introduced a web tool that performs parsing, analysis, and visualization for the entire suite of NVIDIA's genomics computing softwares (Clara Parabricks)}
- Demostrated the usage of the web tool.
- Explained web tool's internal architecture and the role of each components.
- Summarized the technologies used to build and deploy the web application: JavaScript, Python, React, FastAPI, Nginx, Gunicorn.
University of Washington’s Undergraduate Research Symposium
- Intuitively explained the network propagation algorithm: a method that mimic diffusion process to make inference from network data.
- Limitation of network propagation: no confidence estimation for its inference results, which can be limiting for biological discovery.
- Our method: integrating network propagation with the knockoff framework to efficiently estimate confidence level (q-value).
- Showcased experiment results of applying network propagation with confidence estimation on protein homology discovery.