I’m a 1st year Ph.D. student at MIT EECS, recently graduated from National Taiwan University (NTU) with a Bachelor’s degree in Electrical Engineering. I joined the Spoken Language Systems Group at MIT CSAIL in the fall of 2018, working with Dr. James Glass.
I am currently working on various projects on speech and language, including automatic fact-checking and speech synthesis. I am generally interested in learning and analyzing language representations, and transfer learning for NLP.
During my undergrad studies, I researched on speech and language understanding under Prof. Hung-Yi Lee and Prof. Lin-Shan Lee at NTU, and worked on machine learning problems for computer vision under Prof. Yu-Chiang Frank Wang at Academia Sinica/NTU. I also interned twice with the NLP team at Apple.
Here is my Curriculum Vitae (updated Apr 2019).
PhD Student in EECS, 2018 - Present
Massachusetts Institute of Technology
BSE in Electrical Engineering, 2013 - 2018
National Taiwan University
Inspired by the way humans utilize semantic knowledge between objects of interests, we propose a novel deep learning framework that incorporates knowledge graphs for describing the relationships between multiple labels to achieve multi-label zero-shot learning (ML-ZSL).
We built a modular dialogue system with deep learning techniques for the task of assisting people in department stores.
At the 2016 HackNTU Hackathon, our team developed MasterView, a virtual reality system that shows the expert’s moves in first-person perspective so the user can imitate and practice.
Using common and well-known development tools, we transform slither.io, a game based on ‘Snake’, into an interactive augmented-reality game.
In the 2015 CAD Contest, we developed a system to identify equivalent cuts in both equivalent and nonequivalent circuits for design partitioning.
At the 2015 HackNTU Hackathon, our team built a second-hand trading platform, uhu, which provided services for booking, notifications, and filtering functions based on Facebook trading groups, in under 42 hours.