Publications

A relational graph learning approach for robotic crowd navigation using model-based deep reinforcement learning that plans actions by looking into the future.
In Submission to CoRL, 2019

End-to-end framework that learns to model interactions in a data-driven fashion along with a policy to plan ahead.
ICRA, 2019

The first principled treatment of deep network compression under operational constraints.
ECCV, 2018

Experiences

[Jan. 2019 ~ Current] Research Assistant@VML, SFU, Canada
[May 2018 ~ Dec. 2018] Research Intern@VITA, EPFL, Switzerland
[May 2017 ~ April 2018] Research Assistant@VML, SFU, Canada

Talks

Navigation in Crowds: From 2D Navigation to Visual Navigation
Invited Talk at SwissAI Meetup, Lausanne, Switzerland

Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning
Invited Talk at Swiss Machine Learning Day, Lausanne, Switzerland

Projects

Learning to learn sparsity

Use a meta-network to learn importance and correlation of filters and prune the neural network weights accordingly.

constraint-aware neural network compression

The first principled treatment of deep network compression under operational constraints. This framework is both platform independent and constraint type independent.

Reinforced agglomerative clustering

To overcome the greediness of traditional linkage criteria in agglomerative clustering, we proposed a reinforcement learning approach to learn a non-greedy merge policy.

action recognition

Implement several state-of-art models including single frame ResNet, LRCN, optical flow model and two-stream model.

Hornors

President’s & Dean’s Honour Roll (Oct. 2017)
SFU Alumni Scholarship (Sept. 2017)
SFU Alumni Scholarship (Jan. 2017)
Simon Fraser University Entrance Award (Sept. 2016)