CryptoDB
Francisco Rodriguez
Publications
Year
Venue
Title
2024
EUROCRYPT
Polynomial Time Cryptanalytic Extraction of Neural Network Models
Abstract
Billions of dollars and countless GPU hours are currently spent on training Deep Neural Networks (DNNs) for a variety of tasks. Thus, it is essential to determine the difficulty of extracting all the parameters of such neural networks when given access to their black-box implementations. Many versions of this problem have been studied over the last 30 years, and the best current attack on ReLU-based deep neural networks was presented at Crypto'20 by Carlini, Jagielski, and Mironov. It resembles a differential chosen plaintext attack on a cryptosystem, which has a secret key embedded in its black-box implementation and requires a polynomial number of queries but an exponential amount of time (as a function of the number of neurons).
In this paper, we improve this attack by developing several new techniques that enable us to extract with arbitrarily high precision all the real-valued parameters of a ReLU-based DNN using a polynomial number of queries \emph{and} a polynomial amount of time. We demonstrate its practical efficiency by applying it to a full-sized neural network for classifying the CIFAR10 dataset, which has 3072 inputs, 8 hidden layers with 256 neurons each, and about $1.2$ million neuronal parameters. An attack following the approach by Carlini et al.\ requires an exhaustive search over $2^{256}$ possibilities.
Our attack replaces this with our new techniques, which require only 30 minutes on a 256-core computer.
Coauthors
- Isaac Canales-Martínez (1)
- Jorge Chavez-Saab (1)
- Anna Hambitzer (1)
- Francisco Rodriguez (1)
- Nitin Satpute (1)
- Adi Shamir (1)