technology hard Fill in the Blank

The technique that adds noise to model gradients during training to prevent privacy leaks is called ________ SGD.

  1. Hybrid
  2. Differential Privacy / DP
  3. Oxygen / ISRU
  4. bit

Answer: Differential Privacy / DP

DP-SGD adds calibrated noise to gradients during training, providing differential privacy guarantees. Critical for privacy-preserving ML training.

Topic AI/ML Security
Exam Relevance Banking, UPSC, SSC
More technology Questions Back to all questions