Deep learning model watermark embedding method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of information security, and discloses a deep learning model watermark embedding method, which comprises the following steps of: acquiring a watermark signature of a user and a trained deep learning model, and inputting the watermark signature into the deep learning model for training to obtain a training value; calculating a loss function value of thetraining value, comparing the loss function value with a preset threshold value, adjusting the watermark signature according to a comparison result until the loss function value is smaller than the preset threshold value, and outputting the adjusted watermark signature; and performing encoding operation on the adjusted watermark signature by using the deep learning model to obtain a plurality of sub-watermark signatures, and embedding the sub-watermark signatures into different layers of the deep learning model. The invention further provides a deep learning model watermark embedding device, electronic equipment and a computer readable storage medium. According to the deep learning model watermark embedding method and the deep learning model watermark embedding device, property right protection of the deep learning model can be realized.

Description

technical field [0001] The present invention relates to the technical field of information security, in particular to a watermark embedding method, device, electronic equipment and computer-readable storage medium of a deep learning model. Background technique [0002] At present, deep learning models have been widely used in image recognition, classification and other industries. Although the programming framework of the deep learning model makes it easy to use the deep learning model in practical project applications, the training of the deep learning model is still a difficult task because it requires a lot of data and time. Therefore, in the research and development of deep learning models, it is very important to share the trained deep learning models. Sharing will bring about the protection of property rights of deep learning models, and there is no better technology to realize the protection of property rights of deep learning models. Contents of the invention [...

AI Technical Summary

Problems solved by technology

Sharing will bring about the protection of property rights of deep learning models, and there is no better technology to realize the protection of property rights of deep learning models

Images