Time sequence classification method based on LSTM and multi-scale FCN

Abstract

The invention provides a time sequence classification method based on LSTM and multi-scale FCN, belonging to the field of time sequence classification. The method comprises the following steps: setting a general structure of a multi-modal network; extracting time-dependent characteristics by using a long-short-term memory network; fully mining the geometric space characteristics of multiple granularities of a time sequence curve by using a full convolution module; integrating spatio-temporal characteristics and discriminating a sample according to the integrated spatio-temporal characteristics; and fully training a model by using a back propagation algorithm. According to the method, spatial features can be comprehensively explored through a large-scale and multi-scale receptive field, long-term and short-term dependency can be adaptively learned, and learned beneficial information is more comprehensive compared with an existing model. More accurate judgment can be given by more comprehensively mastering the difference characteristics of time sequences.

Description

technical field [0001] The invention relates to the field of time series classification, in particular to a time series classification method based on LSTM and multi-scale FCN. Background technique [0002] In the era of big data, various types of structured and unstructured data can be found everywhere. Time series data, like images and text, is also an extremely common form. It is numerical data obtained by continuously sampling one or more physical quantities at equal time intervals. Time Series Classification (TSC) is a ubiquitous and significant topic. For example, in the industrial field, the pressure exerted on mechanical equipment and the data collected by vibration sensors are time-series data. Through this information, it is possible to determine whether the current part or the whole machine has failed and what kind of failure has occurred, and then give maintenance suggestions; In the medical field, waveform data such as electrocardiograms are also time-series ...

AI Technical Summary

Problems solved by technology

[0009] (1) The attribute dimension is too high (that is, the time step is too long)

The sequence time step of the data set makes the global difference of the sequence less obvious, increasing the difficulty of learning

In addition, too high attribute dimension will lead to high complexity of some methods, which makes it difficult to apply

[0010] (2) The samples are not aligned in time

So the resulting sequence is not perfectly aligned on the time step and will often be misaligned

This makes it impossible to directly use the differences shown by different samples at the same position as the basis for classifying them, and it also brings difficulties to the global similarity between samples.

[0011] The above two defects make it difficult to directly obtain useful information for classification tasks from the original time series

Images