Neural network training and image classification method and device, equipment, medium and program product

Abstract

The embodiment of the invention discloses a neural network training and image classification method and device, equipment, a medium and a program product, and the method comprises the steps: obtaining the training image data of a target task, and the training image data comprises an input sample and a label corresponding to the input sample; processing the input sample through a visual encoder constructed based on a pulse neural network SNN to obtain a feature vector; under the condition of keeping a predetermined semantic vector unchanged, training the visual encoder according to the semantic vector, the feature vector and the tag; the semantic vector represents semantic information of the name of each category corresponding to the target task.

Images