Feature extraction of time series classification based on multi-method integration
| Autor: | Li-Juan Ge, Li Ge |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Series (mathematics)
business.industry Computer science Feature vector Feature extraction Wavelet transform 020206 networking & telecommunications Pattern recognition 02 engineering and technology Atomic and Molecular Physics and Optics Electronic Optical and Magnetic Materials Scale space Wavelet Fractal 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence Electrical and Electronic Engineering Time series business |
| Zdroj: | Optik. 127:11070-11074 |
| ISSN: | 0030-4026 |
| DOI: | 10.1016/j.ijleo.2016.08.089 |
| Popis: | On the basis of analyzing the characteristics of time series data, we propose a feature extraction method of time series classification combining wavelet, fractal and statistic methods. First of all, the original time series is de-noised by using wavelet transform, the de-noised and reconstructed signal is decomposed and the average high frequency coefficients in each scale space are calculated to constitute the feature vectors as the first part of time series classification features; Secondly, we analyze the multi-fractal spectrum of the de-noised and reconstructed signal at multiple scales, and extract the relevant parameters of multi-fractal spectrum as the second part of time series classification features according to the characteristics of specific time series data and classification need; And then according to different characteristics of time series data, extract the relevant statistical characteristics of time series as the third part of time series classification features; Finally, combining the characteristics of time series and experimental results, the extracted features by using wavelet, fractal and statistical methods are analyzed, and the final time series classification features are identified. By comparison with other feature extraction methods, we show the feasibility and superiority of the proposed method using Japanese Vowels data from UCI dataset. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect