Text classification method and system based on dynamic multilayer semantic perceptron

Abstract

The invention provides a text classification method and system based on a dynamic multilayer semantic perceptron. The method comprises the following steps: acquiring a to-be-classified text; adopting a dynamic multilayer semantic perceptron model to obtain a category to which the to-be-classified text belongs; wherein the dynamic multilayer semantic perceptron model comprises a word embedding layer, a plurality of layers of weighted feature learners and a dynamic depth controller, the weighted feature learners are connected in sequence, and each layer of weighted feature learner is composed of a semantic perceptron and a base classifier which are connected in sequence; the output of the word embedding layer is used as the input of the semantic perceptron of all the weighted feature learners, the output of the base classifier of each layer of weighted feature learner is input into the dynamic depth controller, and the output of the semantic perceptron of each layer of weighted feature learner is used as the input of the semantic perceptron of the next layer of weighted feature learner. The feature learning quality is improved, the time complexity and the space complexity are reduced, the model parameter quantity is reduced, and the method is suitable for the situation that training samples are insufficient.

Description

technical field [0001] The invention belongs to the technical field of natural language processing, and in particular relates to a text classification method and system based on a dynamic multi-layer semantic perceptron. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] In the field of Natural Language Processing (NLP), Self-Attention (SA) is a feature weight representation method used to represent the relationship between different words in the same text. Models using Multi Headed Attention (MHA) (such as Transformer, BERT, etc.) have shown excellent performance on a variety of machine learning tasks, and are widely used in text classification, machine translation, named entity recognition, image classification, etc. various fields. However, the shortcomings of the Transformer model using MHA are also obvious: (1) it relies on a large num...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the technical problems in the above-mentioned background technology, the present invention provides a text classification method and system based on a dynamic multi-layer semantic perceptron, which uses a multi-layer semantic perceptron as a feature learner, which improves the quality of feature learning and reduces the time Complexity and space complexity, the dynamic depth controller is used to independently optimize the model depth, reducing the amount of model parameters, which is suitable for solving the problem of deep learning's dependence on large-scale training samples, and the problem of too large and complex models. It provides a feasible method to broaden the practical application of deep learning

Images