Cerebral cortex learning mode-based classification method

Abstract

The invention provides a cerebral cortex learning mode-based classification method which comprises the following steps of: judging whether a training sample set is a continuous data sample or not, if so, performing data discretization processing on the training sample set, constructing an NMN model by using the training sample set, and training the NMN model by using a voting method, and assigning a weight value to a connection line between neural nodes in the NMN model to obtain an NMN classifier, and finally inputting to-be-classified sample data into the NMN classifier to obtain a classification result. According to the method, the algorithm thought of an existing algorithm model is simplified, the problem that the algorithm model needs to be trained by a large number of training sets is solved, and the method has good performance on classification precision of a small data set and a discrete data set.

Description

technical field [0001] The invention relates to the fields of pattern recognition and data classification, in particular to a classification method based on learning patterns of the cerebral cortex. Background technique [0002] At present, the algorithms for simulating the operation mode of the cerebral cortex mainly include the HTM algorithm proposed by Numenta and the Memory Network algorithm proposed by Facebook. These two algorithms are widely used in various fields. HTM has designed a hierarchical memory system designed to simulate the working principle of the neocortex, transforming complex problems into pattern matching and prediction. Each layer of its hierarchical structure symbolizes a cell layer of the real neocortex , its essence can be considered as a memory storage and playback system. The memory network is an algorithm model proposed by Facebook, which abstracts the memory network model into a network model framework. The memory network model framework is m...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the deficiencies in the prior art, the present invention provides a classification method based on the learning mode of the cerebral cortex, which not only simplifies the existing algorithm model algorithm thought , also solves the problem that the algorithm model needs a large number of training sets for training, and this method has a good performance in the classification accuracy of small data sets and discrete data sets

Images