[Topic] Few-shot Learning and Model Adaptation
Despite recent success of deep neural networks, it remains challenging to efficiently learn new concepts from limited training data. To address this problem, our group has been working on several novel few-shot learning and model adaptation strategies, focusing on the structural representation of input data. In particular, we have developed methods for attention-based few-shot image classification, few-shot activity localization in video, and fast-adaptive meta-RL for policy learning.
Learning Context-aware Task Reasoning for Efficient Meta-reinforcement Learning Haozhe Wang, Jiale Zhou, Xuming He, International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2020
A Dual Attention Network with Semantic Embedding for Few-shot Learning Shipeng Yan, Songyang Zhang, Xuming He AAAI Conference on Artificial Intelligence (AAAI),2019
One-shot Action Localization by Learning Sequence Matching Network Hongtao Yang, Xuming He, Fatih Porikli IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018