关于“使用 TensorFlow Privacy 在机器学习中应用差分隐私”的评价

Jinwook G. · 已于 over 1 year前审核

Adnan S. · 已于 over 1 year前审核

Eoin K. · 已于 over 1 year前审核

수빈 박. · 已于 over 1 year前审核

Kyoungeun L. · 已于 over 1 year前审核

지윤 임. · 已于 over 1 year前审核

Alexey S. · 已于 over 1 year前审核

현주 나. · 已于 over 1 year前审核

Hikam M. · 已于 over 1 year前审核

KIHYEON J. · 已于 over 1 year前审核

GEONWOO S. · 已于 over 1 year前审核

motioner D. · 已于 over 1 year前审核

정원 문. · 已于 over 1 year前审核

하 유. · 已于 over 1 year前审核

용성 이. · 已于 over 1 year前审核

Charlie K. · 已于 over 1 year前审核

혁 이. · 已于 over 1 year前审核

Yuchieh Cheng 鄭宇傑 E. · 已于 over 1 year前审核

Leslie M. · 已于 over 1 year前审核

Sohyun K. · 已于 over 1 year前审核

123

Emir S. · 已于 over 1 year前审核

가현 이. · 已于 over 1 year前审核

euiseok l. · 已于 over 1 year前审核

영집 김. · 已于 over 1 year前审核

Introduction to the topic of Privacy Budget was useful. However, the lab would have been more effective if the lab adopted the following approach: 1) Train the model using Privacy Budget n 2) Test the results 3) Retrain the model using Privacy Budget n+delta 4) Test the results 5) Observe the difference in model behaviour between the model using Privacy Budget n vs the model behaviour using Privacy Budget n+delta This would allow the lab user to observe that a lower privacy budget bounds more tightly an adversary's ability to improve their guess.

Paul C. · 已于 over 1 year前审核