Multi-task Self-Supervised Visual Learning
This addresses the challenge of learning unified visual representations without manual labeling, which is incremental as it builds on existing self-supervised methods.
The paper tackled the problem of training a single visual representation by combining multiple self-supervised tasks, showing that deeper networks and task combination improve performance, with their best joint network nearly matching PASCAL detection performance of an ImageNet pre-trained model and matching it on NYU depth prediction.
We investigate methods for combining multiple self-supervised tasks--i.e., supervised tasks where data can be collected without manual labeling--in order to train a single visual representation. First, we provide an apples-to-apples comparison of four different self-supervised tasks using the very deep ResNet-101 architecture. We then combine tasks to jointly train a network. We also explore lasso regularization to encourage the network to factorize the information in its representation, and methods for "harmonizing" network inputs in order to learn a more unified representation. We evaluate all methods on ImageNet classification, PASCAL VOC detection, and NYU depth prediction. Our results show that deeper networks work better, and that combining tasks--even via a naive multi-head architecture--always improves performance. Our best joint network nearly matches the PASCAL performance of a model pre-trained on ImageNet classification, and matches the ImageNet network on NYU depth prediction.