Unsupervised Representation Learning via Neural Activation Coding

Abstract

We present neural activation coding (NAC) as a novel approach for learning deep representations from unlabeled data for downstream applications. We argue that the deep encoder should maximize its nonlinear expressivity on the data for downstream predictors to take full advantage of its representation power. To this end, NAC maximizes the mutual information between activation patterns of the encoder and the data over a noisy communication channel. We show that learning for a noise-robust activation code increases the number of distinct linear regions of ReLU encoders, hence the maximum nonlinear expressivity. More interestingly, NAC learns both continuous and discrete representations of data, which we respectively evaluate on two downstream tasks: (i) linear classification on CIFAR-10 and ImageNet-1K and (ii) nearest neighbor retrieval on CIFAR-10 and FLICKR-25K. Empirical results show that NAC attains better or comparable performance on both tasks over recent baselines including SimCLR and DistillHash. In addition, NAC pretraining provides significant benefits to the training of deep generative models. Our code is available at https://github.com/yookoon/nac.

arXiv commentPublished in International Conference on Machine Learning (ICML), 2021

References in corpus (9)

• Auto-Encoding Variational Bayes
• A Simple Framework for Contrastive Learning of Visual Representations
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• On the Number of Linear Regions of Deep Neural Networks
• Learning Discrete Representations via Information Maximizing Self-Augmented Training
• On Mutual Information Maximization for Representation Learning
• On the number of response regions of deep feed forward networks with piece-wise linear activations
• Inference Suboptimality in Variational Autoencoders
• Iterative Amortized Inference