Auto-Transfer: Learning to Route Transferrable Representations

TL;DR

Auto-Transfer introduces an adversarial multi-armed bandit method to improve transfer learning by automatically routing source representations to target networks, achieving significant accuracy gains especially on small datasets.

Contribution

The paper presents a novel adversarial multi-armed bandit approach for dynamic routing of source features to target networks, enhancing transfer learning effectiveness.

Findings

Achieves over 5% accuracy improvement on benchmark datasets.

More effective for small target datasets.

Qualitative analysis shows better feature focus in target networks.

Abstract

Knowledge transfer between heterogeneous source and target networks and tasks has received a lot of attention in recent times as large amounts of quality labeled data can be difficult to obtain in many applications. Existing approaches typically constrain the target deep neural network (DNN) feature representations to be close to the source DNNs feature representations, which can be limiting. We, in this paper, propose a novel adversarial multi-armed bandit approach that automatically learns to route source representations to appropriate target representations following which they are combined in meaningful ways to produce accurate target models. We see upwards of 5\% accuracy improvements compared with the state-of-the-art knowledge transfer methods on four benchmark (target) image datasets CUB200, Stanford Dogs, MIT67, and Stanford40 where the source dataset is ImageNet. We qualitatively analyze the goodness of our transfer scheme by showing individual examples of the important features focused on by our target network at different layers compared with the (closest) competitors. We also observe that our improvement over other methods is higher for smaller target datasets making it an effective tool for small data applications that may benefit from transfer learning.