Learning More Universal Representations for Transfer-Learning

TL;DR

This paper introduces methods to enhance the universality of visual representations for transfer learning, focusing on reducing reliance on annotated data while demonstrating effectiveness across diverse visual tasks.

Contribution

It proposes two novel approaches to improve universality with minimal annotated data and introduces a unified framework based on diversifying training problems.

Findings

Effective improvement in universality across 10 visual domains

Reduces need for annotated data in training

Demonstrates benefits on classification and diverse visual tasks

Abstract

A representation is supposed universal if it encodes any element of the visual world (e.g., objects, scenes) in any configuration (e.g., scale, context). While not expecting pure universal representations, the goal in the literature is to improve the universality level, starting from a representation with a certain level. To do so, the state-of-the-art consists in learning CNN-based representations on a diversified training problem (e.g., ImageNet modified by adding annotated data). While it effectively increases universality, such approach still requires a large amount of efforts to satisfy the needs in annotated data. In this work, we propose two methods to improve universality, but pay special attention to limit the need of annotated data. We also propose a unified framework of the methods based on the diversifying of the training problem. Finally, to better match Atkinson's cognitive study about universal human representations, we proposed to rely on the transfer-learning scheme as well as a new metric to evaluate universality. This latter, aims us to demonstrates the interest of our methods on 10 target-problems, relating to the classification task and a variety of visual domains.