Research

In high school, I was obsessed with geometry and did a full survey on generalised Fermat problem. I then further studied its related NP-Hard Euclidean Steiner Tree problem during my undergrad study.

Currently, I am interested in machine learning, in particular meta-learning, self-supervised learning and multi-task learning with applications in vision and robotics. The following is my publications displayed in chronological order.

Thesis

• Master Thesis: Universal Representations: Towards Multi-task Learning & Beyond.
  *Awarded as distinguished thesis, Department of Computing at Imperial College.

Publication

• Shape Adaptor: A Learnable Resizing Module
  European Conference on Computer Vision (ECCV), 2020
  Shikun Liu, Zhe Lin, Yilin Wang, Jianming Zhang, Federico Perazzi, and Edward Johns
  [paper] [arxiv] [code] [slide]
• Self-Supervised Generalisation with Meta Auxiliary Learning
  Advances in Neural Information Processing Systems (NeurIPS), 2019
  Shikun Liu, Andrew J. Davison and Edward Johns
  [paper] [arxiv] [code] [poster] [slide]
• End-to-End Multi-task Learning with Attention
  Computer Vision and Pattern Recognition (CVPR), 2019
  Shikun Liu, Edward Johns, and Andrew J. Davison
  [paper] [arxiv] [code]
• Learning a Hierarchical Latent-Variable Model of 3D Shapes
  International Conference on 3D Vision (3DV), 2018 [ORAL]
  Shikun Liu, C. Lee Giles, and Alexander G. Ororbia II.
  [paper] [supplementary] [arxiv] [code] [project page]