Curriculum Learning for Deep Generative Models with Clustering

TL;DR

This paper introduces a cluster-based curriculum learning algorithm for training deep generative models on noisy data, improving model quality by prioritizing central data clusters and ensuring scalability.

Contribution

It proposes a novel clustering-based curriculum learning method with an active set strategy for scalable training of generative models on noisy datasets.

Findings

The algorithm effectively learns high-quality generative models on noisy data.

The optimal cluster curriculum relates to the critical point of geometric percolation.

Experimental results validate improved model performance on face datasets.

Abstract

Training generative models like Generative Adversarial Network (GAN) is challenging for noisy data. A novel curriculum learning algorithm pertaining to clustering is proposed to address this issue in this paper. The curriculum construction is based on the centrality of underlying clusters in data points. The data points of high centrality takes priority of being fed into generative models during training. To make our algorithm scalable to large-scale data, the active set is devised, in the sense that every round of training proceeds only on an active subset containing a small fraction of already trained data and the incremental data of lower centrality. Moreover, the geometric analysis is presented to interpret the necessity of cluster curriculum for generative models. The experiments on cat and human-face data validate that our algorithm is able to learn the optimal generative models (e.g. ProGAN) with respect to specified quality metrics for noisy data. An interesting finding is that the optimal cluster curriculum is closely related to the critical point of the geometric percolation process formulated in the paper.