Drug classification method based on self-encoding and extreme learning machine

Abstract

The invention provides a dual-band spectrum identification method (DWAE-ELM) based on an auto-encoding network and an extreme learning machine aiming at the classification problem of near infrared spectrum data of medicines, the method combines the advantages of an AE method and an ELM method, the AE is used for extracting two-dimensional characteristics of the near infrared spectrum data of the medicines, and the ELM is used for classification according to the characteristics. The DWAE-ELM network is structurally divided into two independent stages: in the first stage, a three-layer AE network is adopted to extract sparse features of two-dimensional input data after dual-band transformation to perform unsupervised multi-level feature representation; and in the second stage, the original ELM is used for performing a final medicine classification task. According to the method, the advantages of high feature extraction capability of the self-encoding network and high ELM training speed are combined, the accuracy and stability of drug classification are improved, and compared with other methods, the model training time is greatly shortened, the method is not sensitive to the size of a training set, and the robustness is higher.

Description

Technical field [0001] The present invention relates to a pharmaceutical classification method, and more particularly to a pharmaceutical classification method based on the self-proder and ultimate learning machine. Background technique [0002] At present, although deep learning has been widely used in image, voice, text and other fields, it has made good results, but the application on the near-infrared spectrum is still less, because near-infrared nature is one-dimensional vector, and the data set tends to Not big, although deep learning has strong learning ability, it is easy to fit, and the traditional depth learning network structure is not suitable for processing one-dimensional data. At present, near-infrared spectrum classification is mainly used by machine learning, such as reverse propagation algorithm (BP), support vector machine (SVM), limit learning machine, etc., and these algorithms have shown a relatively powerful performance. However, these machine learning algo...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of unstable classification performance of traditional ELM and the high dimensionality of raw drug data, irrelevant information will affect the classification performance, the present invention combines the dual-band transformation method, combines the autoencoder network and the extreme learning machine, and proposes a A New Drug Classification Method Based on DWAE-ELM

Images