Data processing method, device and system and storage medium

Abstract

The invention provides a data processing method, device and system and a storage medium. The method comprises the steps of obtaining to-be-processed data; performing feature extraction on the to-be-processed data to obtain data features of the to-be-processed data; inputting the data features into a trained data processing model to obtain a data processing result; wherein the training of the data processing model comprises the following steps: selecting a number type of training samples from training data; inputting the category weight of the number category into a layer of a corresponding category of the neural network; training the neural network by adopting the training samples to obtain new category weights of the quantity categories; and repeating the above steps, and stopping training when the neural network model satisfies a training condition to obtain a trained data processing model. According to the method, the device, the system and the storage medium, the neural network model obtained by training the neural network by disorderly selecting the training samples with the fixed number of categories is easy to converge, the accuracy is high, the method and the device are suitable for data processing application, and the data processing process is optimized.

Description

Technical field [0001] The present invention relates to the field of data processing technology, and more specifically to data processing methods, devices, systems and storage media. Background technique [0002] With the widespread application of neural networks in image processing, text processing, and language processing, the demands and requirements for neural networks have gradually increased. Neural networks often need to contain category information to complete multi-category tasks such as classification, detection of different categories, or regression. Usually, because these category information is also undetermined, it needs to be optimized or optimized during the training process. These category information are often stored as network weights in a specific layer of the neural network to participate in training and inference. The more categories there are, the more weights are required, and the slower the training and inference of these layers. [0003] At present, when...

AI Technical Summary

Problems solved by technology

The present invention provides a data processing method, device, system and computer storage medium to solve the problem that the neural network model is difficult to converge, requires a long training time, and cannot train category weights, which is not conducive to the optimization of the data processing process

Images