MLSys Seminar: General Visual Representation Learning with Multi-Task Learning
Abstract: In this talk, I will introduce and discuss various topics centered around multi-task learning designed for computer vision tasks, including multi-task optimization strategies, task relationships, and multi-task architecture designs. I will share my intuition and behind-the-paper thinkings in developing MAXL, Auto-Lambda, and Prismer, which are three projects I have been worked on during my Ph.D. Finally, I will offer a glimpse into my personal perspectives and engage in open-ended discussions concerning the future trajectory of multi-task learning research.
Bio: Shikun Liu is a fourth-year PhD student at Dyson Robotics Lab in Imperial College, co-advised by Prof. Andrew Davison and Prof. Edward Johns. Shikun's main research goal is to develop general-purpose multi-task and multi-modal learning systems. To that end, his work has broadly concerned with the study of multi-task relationships, multi-task and auxiliary learning method design, and self and semi-supervised learning frameworks.
Please contact lijunzhang [at] cs.umass.edu (Lijun Zhang) with any questions.