Deep belief network image recognition method based on Bayesian regularization

Abstract

The invention discloses a deep belief network image recognition method based on Bayesian regularization and belongs to the field of artificial intelligence and machine learning. The deep belief network plays a more and more important role in the field of digital detection and image recognition. The invention provides a deep belief network based on Bayesian regularization on the basis of the network sparsity characteristic and changes of connection weights to solve the problem of overfitting in the training process of the deep belief network. By applying Bayesian regularization to the network training process, balance between error decreasing and weight increasing is effectively adjusted. The classification experiment of a digital script database proves effectiveness of the improved algorithm. An experimental result shows that in the deep belief network, the deep belief network image recognition method can effectively overcome the overfitting phenomenon and improve accuracy of digital recognition.

Description

technical field [0001] The invention utilizes a Bayesian Regularization-based deep learning network (Deep BeliefNetwork, DBN) to realize recognition and classification of handwritten digital images. Neural network is an important method in the field of artificial intelligence and neural computing, and image detection and recognition is one of the important tasks in the field of machine learning, so the present invention belongs to the fields of artificial intelligence and machine learning. Background technique [0002] Digital identification technology has a place in many industries, such as education, transportation, commerce, postal services and banking. The realization and application of automatic identification of numbers provides important convenience for people's life, and is closely related to people's life. It is widely used in number detection of vehicles and roads, automatic identification of personal transcripts, etc. Compared with printed digit recognition, hand...

AI Technical Summary

Problems solved by technology

More and more works have shown that the correct setting of network parameters will greatly affect the results

[0004] Overfitting is one of the common problems in the training process of neural networks. In order to improve the generalization ability of the network, traditional methods include simplifying the network structure and stopping training early. These methods work to varying degrees, but are not easy to generalize.

Images