High-dimensional multi-label data flow classification method based on online sequence kernel extreme learning machine

Abstract

The invention discloses a multi-label text data stream classification method based on an online sequence kernel extreme learning machine. The method comprises the steps that 1, constructing a BoW model and a sliding window mechanism according to an external corpus to divide multi-label text data streams into data blocks and then vectorize the data blocks; 2, predicting the text data block Dk at the moment k by utilizing the integrated classifier model at the moment k1, and outputting a prediction result; 3, performing feature selection on the text feature set of the text data block Dk to obtain a dimension-reduced text feature set Mk; 4, judging whether concept drift or feature drift occurs or not according to the cosine similarity between the class label spaces of the text data block Dk at the moment k and the text data block Dk1 at the moment k1 and the distribution difference between the feature sets after dimension reduction; 5, constructing an online sequence kernel extreme learning machine by utilizing all texts in the text data block Dk according to the drift detection condition, and updating to an integrated classifier model at the moment k. According to the method, the classification problem of the multi-label text data flow with feature drift and concept drift is solved.

Description

technical field [0001] The invention belongs to the field of multi-label data flow mining in practical applications, and in particular relates to the multi-label data flow classification problem of text data flow, feature drift and concept drift. Background technique [0002] In many practical application fields, such as: e-mail classification, news feed, medical diagnosis, image recognition, etc., a massive, continuous, high-speed, dynamic data-data flow is generated. These data flow data are widely used in social production and life practice, and have great research value. The data stream type data in the real world usually has the following characteristics: (1) The data generation speed is fast and the scale is massive. According to the statistics of Twitter in 2012, the daily average number of Twitter users exceeds 200 million, and the total number of Tweets sent every day reaches 400 million, with an average of more than 5,000 Tweets per second; (2) An instance object ...

AI Technical Summary

Problems solved by technology

[0003] One of the challenges: the existing multi-label classification algorithm and multi-label feature dimensionality reduction algorithm are both batch processing algorithms with high time complexity

However, the text data stream in the practical application field usually comes continuously, and it is difficult to obtain all the data at one time; and it does not take into account the high-dimensional and dynamic features of the newly arrived text data stream. Therefore, the existing multi-label classification and feature reduction Dimensional algorithms are difficult to be directly used to process massive high-dimensional multi-labeled text data streams

[0004] Challenge 2: It is difficult to detect the feature drift and concept drift caused by the hidden features and label distribution changes in the text data stream, which will easily lead to low accuracy of the classification algorithm; the existing multi-label classification algorithm is also difficult to adapt to the text The characteristics of the data stream and the problem of concept drift have achieved better classification results

Images