Model data processing method based on DNN, data processing model and electronic device

Abstract

The invention provides a Model data processing method based on DNN. The method comprises the steps of providing feature data; Performing DNN processing on the feature data; Performing DNDF processingon at least part of neurons in at least one neuron layer in the DNN processing to obtain output data; And obtaining a prediction result corresponding to the feature data at least according to the output data. The invention further provides a data processing model based on the DNN. The data processing model comprises an input module; Wherein the input module is used for inputting feature data; A DNN processing module; Wherein the DNN module is used for performing DNN processing on feature data; The DNDF processing module is used for performing DNDF processing on at least part of neurons in at least one neuron layer in the DNN processing module to obtain output data; And a prediction module; And the prediction module is used for obtaining a prediction result corresponding to the feature dataat least according to the output data output by the DNDF processing module. The invention further provides an electronic device for executing the model data processing method. The model data processing method based on the DNN, the data processing model and the electronic device have the advantages that prediction precision is improved and the like.

Description

【Technical Field】 [0001] The invention relates to the field of artificial intelligence, and in particular to a DNN-based model data processing method, data processing model and electronic device. 【Background technique】 [0002] Based on artificial neural networks, especially deep neural networks (Deep Neural Networks, hereinafter referred to as DNN processing) have achieved great success in many applications. Especially for processing structured data, DNN processing performance is excellent. Therefore, DNN processing is widely used in the field of machine learning. However, DNN processing also has certain shortcomings. For example, DNN processing in the processing of structured data machine learning problems, the characteristic of fully connected causes the model to be greatly affected by noise. [Content of the invention] [0003] In order to overcome the disadvantages of current DNN processing, the present invention proposes a DNN-based model data processing method, data proces...

AI Technical Summary

Problems solved by technology

However, DNN processing also has certain shortcomings. For example, in the processing of machine learning problems with structured data, the fully connected characteristics of DNN processing cause the model to be greatly affected by noise.

Images