Reconstructing Training Data from Model Gradient, Provably
원문 초록
Understanding when and how much a model gradient leaks information about the training sample is an important question in privacy. In this paper, we present a surprising result: even without training or memorizing the data, we can fully reconstruct the training samples from a single gradient query at a randomly chosen parameter value. We prove the identifiability of the training data under mild conditions: with shallow or deep neural networks and a wide range of activation functions. We also present a statistically and computationally efficient algorithm based on tensor decomposition to reconstruct the training data. As a provable attack that reveals sensitive training data, our findings suggest potential severe threats to privacy, especially in federated learning.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Model Parallelism With Subnetwork Data Parallelism
Private Model Personalization Revisited
Few-Shot Model Agnostic Federated Learning
Federated Model Search via Reinforcement Learning
Making Language Model a Hierarchical Classifier and Generator
FLsim: A Modular and Library-Agnostic Simulation Framework for Federated Learning
토론
이 논문과 연결된 포럼 글이 아직 없습니다.