Data Augmentation and Regularization for Learning Group Equivariance
This work addresses the problem of improving model performance in machine learning tasks with known symmetries, which is relevant to researchers and practitioners working on tasks with inherent symmetries.
The authors tackled the problem of learning group equivariance in machine learning tasks, achieving equivariance of the trained model through training on augmented data and regularization. The exact performance gain is not specified in the abstract.
In many machine learning tasks, known symmetries can be used as an inductive bias to improve model performance. In this paper, we consider learning group equivariance through training with data augmentation. We summarize results from a previous paper of our own, and extend the results to show that equivariance of the trained model can be achieved through training on augmented data in tandem with regularization.