Research Interests:

Reinforcement Learning, Meta-Learning, Curriculum Learning, Multi-task Learning, Sequential Decision Making, Graphical Models, Interpretable ML. Applications in healthcare, biology, medicine, natural language and general AI.


1 Haowen Xu, Hao Zhang, Zhiting Hu, Xiaodan Liang, Ruslan Salakhutdinov, Eric Xing. AutoLoss: Learning Discrete Schedule for Alternate Optimization. ICLR 2019

Research Experience

Research Assistant, [Machine Learning Group], Georgia Institute of Technology, 2020.1-present. Advisor: Prof. Song Le. I am working on end2end protein sequence alignment via deep leanring.

Research Assistant, SAILING LAB, Carnegie Mellon University, 2018.3-2018.7. Advisor: Prof. Eric Xing.
I proposed a meta-learning framework, AutoLoss, that automatically learns and determines the schedule of optimization processes, which can improve the convergence of iterative and alternate training such as GAN, multi-task learning and curriculum learning. See more.

Algorithm Engineer,, Beijing, 2017.9-2018.2
I developed an attention based sequence labeling model and applied it to a Named Entity Recognition task. See more.
I designed a reasoning module to handle the feature-conflict problem in document classification tasks and applied hierarchical supervision to exploit label information of different granularities. See more.

Research Intern, Laboratory of Auditory Neurophysiology, Johns Hopkins University, 2016.8-2017.5
I developed an automatic recording and analyzing system for animal vocalization behavior study and applied deep learning methods to our analyzing algorithm. I tackled the commom engineering problems (e.g. insufficient data, unstable recording system, big individual variance) when applied deep learning to biomedical areas.


I am currently a Master student in CS&E in Georgia Tech. I completed my bachelor degree in Tsinghua University in 2017. In my thesis, I developed and applied deep machine learning methods to biomedical applications. See more about my research in the Projects page.

Other materials:

Word Embedding Review