Task processing method and device based on domain adaptation, electronic equipment and storage medium

The invention provides a task processing method and device based on domain adaptation, electronic equipment and a storage medium, and the method comprises the steps: obtaining to-be-processed data, and inputting the obtained to-be-processed data into a trained deep learning model according to batch; for any batch of normalization layers of the trained deep learning model, determining a distribution statistical value of input data of the current batch by using the batch of normalization layers, and performing batch normalization processing on the input data of the current batch according to the determined distribution statistical value and the trained scale scaling parameter and offset of the batch of normalization layers; and determining a task processing result of the to-be-processed data of the current batch according to an output result of the trained deep learning model for the to-be-processed data of the current batch. According to the method, the accuracy of task processing by using the trained deep lea