An Unknown Pattern Detection Method for Time Series Data Based on Convolutional Neural Network

Abstract

Exploration on the time series data in unknown model pattern recognition has important research significance. This paper proposes an unknown pattern detection method for time-series data based on convolution neural network, which planifies the output results by transforming fully connection layer and softmax layer of the traditional convolutional neural network, and uses the coordinate point and Euclidean distance to determine whether the timing series data belongs to the known pattern or the unknown pattern. Experiments show that the method in this paper can effectively detect the time-series data of unknown patterns and has certain accuracy.