Generative Adversarial Residual Pairwise Networks for One Shot Learning

TL;DR

This paper introduces a novel residual pairwise network with a learnable similarity measure and GAN-based regularization, significantly improving one shot learning performance on mini-ImageNet.

Contribution

It proposes a new residual pairwise network architecture with a learnable similarity function and GAN-based regularization for enhanced one shot learning.

Findings

Outperforms previous state-of-the-art on mini-ImageNet for one shot learning.

Achieves over 55% accuracy in 5-way classification.

Demonstrates effective generation of plausible class variations.

Abstract

Deep neural networks achieve unprecedented performance levels over many tasks and scale well with large quantities of data, but performance in the low-data regime and tasks like one shot learning still lags behind. While recent work suggests many hypotheses from better optimization to more complicated network structures, in this work we hypothesize that having a learnable and more expressive similarity objective is an essential missing component. Towards overcoming that, we propose a network design inspired by deep residual networks that allows the efficient computation of this more expressive pairwise similarity objective. Further, we argue that regularization is key in learning with small amounts of data, and propose an additional generator network based on the Generative Adversarial Networks where the discriminator is our residual pairwise network. This provides a strong regularizer by leveraging the generated data samples. The proposed model can generate plausible variations of exemplars over unseen classes and outperforms strong discriminative baselines for few shot classification tasks. Notably, our residual pairwise network design outperforms previous state-of-theart on the challenging mini-Imagenet dataset for one shot learning by getting over 55% accuracy for the 5-way classification task over unseen classes.