Computer Vision Interview Questions, issue 13, Jan 14, 2026
The Generalization Gap Trap
Senior Computer Vision Engineer interview at Google DeepMind, and the interviewer asks:
โWe use heavy data augmentation (Color Jitter, 30ยฐ Rotations) during training to improve robustness. Why do we strictly disable these during validation? Doesnโt that break the rule that ๐๐ณ๐ข๐ช๐ฏ ๐ข๐ฏ๐ฅ ๐๐ฆ๐ด๐ต ๐ฅ๐ช๐ด๐ต๐ณ๐ช๐ฃ๐ถ๐ต๐ช๐ฐ๐ฏ๐ด ๐ด๐ฉ๐ฐ๐ถ๐ญ๐ฅ ๐ฎ๐ข๐ต๐ค๐ฉ?โ
Why disabling data augmentation during evaluation is the only way to measure real generalization.
The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.