(ML)
Increase the amount of data by adding
- slightly modified copies of already existing data
- newly created synthetic data from existing data.
It’s especially used in image processing or speech recognition to increase the training set size significantly by distorting training examples. However, these distortions MUST be representative in the test set