Learning Broken Symmetries with Approximate Invariance

Seth Nabat, Aishik Ghosh, Edmund Witkowski +2

Recognizing symmetries in data allows for significant boosts in neural network training, which is especially important where training data are limited. In many cases, however, the…