Artificially creating more training data by modifying existing examples. Like teaching someone to recognize cats by showing them the same photo of a cat, but flipped, rotated, colored differently, or wearing a tiny hat.
Synonyms:
Training Data Expansion