Dynamic Few-Shot Visual Learning without Forgetting

TL;DR

This paper presents a novel few-shot visual learning system that efficiently learns new categories with minimal data while retaining knowledge of initial categories, using attention mechanisms and cosine similarity-based classifiers.

Contribution

It introduces an attention-based weight generator and a cosine similarity classifier redesign, enabling simultaneous learning of new and existing categories without forgetting.

Findings

Achieved 56.20% on 1-shot Mini-ImageNet

Achieved 73.00% on 5-shot Mini-ImageNet

Outperformed prior state-of-the-art methods

Abstract

The human visual system has the remarkably ability to be able to effortlessly learn novel concepts from only a few examples. Mimicking the same behavior on machine learning vision systems is an interesting and very challenging research problem with many practical advantages on real world vision applications. In this context, the goal of our work is to devise a few-shot visual learning system that during test time it will be able to efficiently learn novel categories from only a few training data while at the same time it will not forget the initial categories on which it was trained (here called base categories). To achieve that goal we propose (a) to extend an object recognition system with an attention based few-shot classification weight generator, and (b) to redesign the classifier of a ConvNet model as the cosine similarity function between feature representations and classification weight vectors. The latter, apart from unifying the recognition of both novel and base categories, it also leads to feature representations that generalize better on "unseen" categories. We extensively evaluate our approach on Mini-ImageNet where we manage to improve the prior state-of-the-art on few-shot recognition (i.e., we achieve 56.20% and 73.00% on the 1-shot and 5-shot settings respectively) while at the same time we do not sacrifice any accuracy on the base categories, which is a characteristic that most prior approaches lack. Finally, we apply our approach on the recently introduced few-shot benchmark of Bharath and Girshick [4] where we also achieve state-of-the-art results. The code and models of our paper will be published on: https://github.com/gidariss/FewShotWithoutForgetting