Neural network model reference neural cognitive mechanism and machine learning mathematical method

Abstract

The invention provides a neural network model reference neural cognitive mechanism and machine learning mathematical method, and relates to the field of machine learning. The method comprises the following steps: S1, establishing a sample library for sample information, storing the sample library, generating adversarial samples through a fast gradient symbol method, and dividing the adversarial samples into a training set and a test set; S2, enhancing the migration ability from the adversarial sample library to the training set through lifelong continuous migration learning; S3, connecting a recurrent neural network through a convolutional neural network, and then connecting a BP neural network and a naive Bayes algorithm to establish a neural network model; and S4, classifying the target information in the adversarial sample library to form a data table. The adversarial samples are generated through the fast gradient symbol method, videos, images and texts are combined, real scene dialogues, dialogues with images and text expression are more specific, the artificial intelligence application range is widened, the observation thinking and expression ability is improved, and generalization is improved through different structures.

Description

technical field [0001] The invention relates to the field of machine learning, specifically, the neural network model uses neurocognitive mechanism and machine learning mathematical method for reference. Background technique [0002] Since the advent of the mathematical method of simulating the actual human neural network, people have gradually become accustomed to directly calling this artificial neural network a neural network. Neural network has broad and attractive prospects in the fields of system identification, pattern recognition, intelligent control, etc. Especially in intelligent control, people are particularly interested in the self-learning function of neural network, and regard this important feature of neural network as a It is one of the keys to solve the difficult problem of controller adaptability in automatic control. [0003] Deep neural network (DNN) is a multi-layer perceptron architecture used to solve complex learning problems. However, DNN faces cha...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies of the prior art, the present invention provides a neural network model that draws on neurocognitive mechanisms and machine learning mathematical methods, and solves the problem that a large amount of data interconnected by DNN may be overfitting

Images