Model watermark construction method and device, equipment and storage medium

The invention provides a model watermark construction method and device, electronic equipment and a storage medium, and relates to the field of artificial intelligence, in particular to the technical field of safety of an artificial intelligence network model. According to the specific implementation scheme, a target sample is selected from a sample set of a to-be-protected network model; attacking the target sample to obtain disturbance of successful attack; adding the disturbance which is successfully attacked to the target sample to obtain an adversarial sample; and adjusting parameters of a target neural network layer of the to-be-protected network model based on the adversarial sample to obtain a target model with a watermark. In the embodiment of the invention, the target sample is firstly selected and then attacked so as to obtain appropriate disturbance, and the whole process is short and low in complexity. Besides, the parameters of the target neural network layer are adjusted instead of the paramete