Multi-directional Knowledge Transfer for Few-Shot Learning
Shuo Wang, Xinyu Zhang, Yanbin Hao, Chengbing Wang, Xiangnan He · Proceedings of the 30th ACM International Conference on Multimedia · 2022
Knowledge transfer-based few-shot learning (FSL) aims at improving the recognition ability of a novel object under limited training samples by transferring relevant potential knowledge from other data. Most related methods calculate such knowledge to refine the representation of a novel sample or enrich the supervision to a classifier during a transfer procedure. However, it is easy to introduce new noise during the transfer calculations since: (1) the unbalanced quantity of samples between the known (base) and the novel categories biases the contents capturing of the novel objects, and (2) the semantic gaps existing in different modalities weakens the knowledge interaction during the training.