paper-with-me

홈 › Papers

Opacus: User-Friendly Differential Privacy Library in PyTorch

2021-09-25 · Ashkan Yousefpour, Igor Shilov, Alexandre Sablayrolles, Davide Testuggine, Karthik Prasad, Mani Malek, John Nguyen, Sayan Ghosh, Akash Bharadwaj, Jessica Zhao, Graham Cormode, Ilya Mironov

We introduce Opacus, a free, open-source PyTorch library for training deep learning models with differential privacy (hosted at opacus.ai). Opacus is designed for simplicity, flexibility, and speed. It provides a simple and user-friendly API, and enables machine learning practitioners to make a training pipeline private by adding as little as two lines to their code. It supports a wide variety of layers, including multi-head attention, convolution, LSTM, GRU (and generic RNN), and embedding, right out of the box and provides the means for supporting other user-defined layers. Opacus computes batched per-sample gradients, providing higher efficiency compared to the traditional "micro batch" approach. In this paper we present Opacus, detail the principles that drove its implementation and unique features, and benchmark it against other frameworks for training models with differential privacy as well as standard PyTorch.

📄 PDF Abstract BibTeX arXiv:2109.12298

Code (3)

pytorch/opacus 공식 구현 pytorch
alexbie98/dpgan-revisit pytorch
woodyx218/private_vision pytorch

Methods 이 논문이 사용한 방법론

GRU A Gated Recurrent Unit, or GRU, is a type of recurrent neural network. It is similar to an LSTM, but only has two gates - a reset…

Similar Papers 제목 키워드 기반

NanoBatch Privacy: Enabling fast Differentially Private learning on the IPU

2021-09-24 · Edward H. Lee, Mario Michael Krell, Alexander Tsyplikhin, Victoria Rege 외

Differentially private SGD (DPSGD) has recently shown promise in deep learning. However, compared to non-private SGD, the DPSGD algorithm places computational overheads that can undo the benefit of batching in GPUs. Micr…

Diffprivlib: The IBM Differential Privacy Library

2019-07-04 · Naoise Holohan, Stefano Braghin, Pól Mac Aonghusa, Killian Levacher

Since its conception in 2006, differential privacy has emerged as the de-facto standard in data privacy, owing to its robust mathematical guarantees, generalised applicability and rich body of literature. Over the years,…

FlashDP: Private Training Large Language Models with Efficient DP-SGD

2025-07-01 · Liangyu Wang, Junxiao Wang, Jie Ren, Zihang Xiang 외 arxiv

As large language models (LLMs) increasingly underpin technological advancements, the privacy of their training data emerges as a critical concern. Differential Privacy (DP) serves as a rigorous mechanism to protect this…

Scalable and Efficient Training of Large Convolutional Neural Networks with Differential Privacy

2022-05-21 · Zhiqi Bu, Jialin Mao, Shiyun Xu

Large convolutional neural networks (CNN) can be difficult to train in the differentially private (DP) regime, since the optimization algorithms require a computationally expensive operation, known as the per-sample grad…

On the Convergence and Calibration of Deep Learning with Differential Privacy

2021-06-15 · Zhiqi Bu, Hua Wang, Zongyu Dai, Qi Long

Differentially private (DP) training preserves the data privacy usually at the cost of slower convergence (and thus lower accuracy), as well as more severe mis-calibration than its non-private counterpart. To analyze the…