Method for mining similarity information of time series data based on integrated model

Abstract

The invention discloses a method for mining similarity information of time series data based on an integrated model. The method comprises a hidden Markov model and a conditional variation auto-encoder model based on a Wasserstein distance. The method comprises the following steps: establishing an input layer, and primarily processing an input time sequence; then learning and classifying the input data by a hidden Markov classification layer and a conditional variation encoder layer respectively; after learning is finished, through further optimization, two classification models of two layers are fused through a Stacking algorithm, and parallel training can be carried out. And meanwhile, the Wasserstein distance is innovatively used for replacing KL divergence to measure the distance between the two time sequences, so that the classifier has wider application. According to the method, similar information mining from the hidden state and distribution of the time sequence can be better carried out, and all the mined information can be fused, so that the learning of the model is more effective, the operation efficiency is higher, and the method has wider applicability.

Description

technical field [0001] The invention belongs to the technical fields of data mining and machine learning, and in particular relates to a method for mining similarity information of time series data based on an integrated model. Background technique [0002] In time series data mining, similarity information is the key information, and it is also one of the starting points of data mining. But now in the mining of time series data, many algorithms have lost the similarity information of data distribution, and only calculate the similarity from the perspective of data. This kind of similarity mining that only relies on the data angle is a kind of information loss. This loss will cause some features implicitly included in the time series data to be lost, which will affect the effect of learning, resulting in a large gap between the learned distribution and the real distribution. big. At present, there is a lack of algorithms that utilize time series distribution information. D...

AI Technical Summary

Problems solved by technology

But now in the mining of time series data, many algorithms have lost the similarity information of data distribution, and only calculate the similarity from the perspective of data

This kind of similarity mining that only relies on the data angle is a kind of information loss. This loss will cause some features implicitly included in the time series data to be lost, which will affect the effect of learning, resulting in a large gap between the learned distribution and the real distribution. Big

At present, there is a lack of algorithms that utilize time series distribution information. Distribution similarity is one of the key research issues in statistics. However, in mining time series data information, mining distribution similarity has not been widely discussed.

Images