DP-Forward: Fine-tuning and Inference on Language Models with Differential Privacy in Forward Pass
Minxin Du, Xiang Yue, Sherman S. M. Chow, Tianhao Wang, Chenyu Huang, Huan Sun
Abstract
Differentially private stochastic gradient descent (DP-SGD) adds noise to gradients in back-propagation, safeguarding training data from privacy leakage, particularly membership inference. It fails to cover (inference-time) threats like embedding inversion and sensitive attribute inference. It is also costly in storage and computation when used to fine-tune large pre-trained language models (LMs).
Topics & Concepts
InferenceEmbeddingComputer scienceDifferential privacyStochastic gradient descentAlgorithmGradient descentApproximate inferenceComputationNoise (video)GaussianArtificial intelligenceQuantum mechanicsArtificial neural networkPhysicsImage (mathematics)Privacy-Preserving Technologies in DataAdversarial Robustness in Machine LearningStochastic Gradient Optimization Techniques