Contrastive Representation Learning with Trainable Augmentation Channel
This addresses a specific issue in representation learning for computer vision, but it is an incremental improvement.
The paper tackles the problem of collapsed representations in contrastive learning when augmentations damage image information, by formalizing a stochastic encoding process with an infoMax objective to learn data-dependent augmentation distributions.
In contrastive representation learning, data representation is trained so that it can classify the image instances even when the images are altered by augmentations. However, depending on the datasets, some augmentations can damage the information of the images beyond recognition, and such augmentations can result in collapsed representations. We present a partial solution to this problem by formalizing a stochastic encoding process in which there exist a tug-of-war between the data corruption introduced by the augmentations and the information preserved by the encoder. We show that, with the infoMax objective based on this framework, we can learn a data-dependent distribution of augmentations to avoid the collapse of the representation.