AI RESEARCH

[D] Thinking about augmentation as invariance assumptions

r/MachineLearning

Data augmentation is still used much heuristically than it should be. A