Data-Security-Research-Lab / CNN-prediction-ZKP-scheme

The code corresponds to the paper “Validating the integrity of Convolutional Neural Network predictions based on Zero-Knowledge Proof“

CNN-prediction-ZKP-scheme

The code corresponds to the paper “Validating the integrity of Convolutional Neural Network predictions based on Zero-Knowledge Proof“

Overview

• This library implements a zero-knowledge scheme for proving/verifying the integrity of CNN predictions.

Build instructions

1. Please follow libsnark's instructions to build a zero-knowledge proof runtime environment and install the required core libraries. https://github.com/scipr-lab/libsnark
2. The src folder contains the example CNN model prediction integrity circuit we designed.
3. The sample code for extracting the computational information of the CNN model prediction process is shown in File Model_Extract.py .