Feature-based classification of time-series data

By Alexandros Nanopoulos, R. Alcock, Yannis Manolopoulos. 2001

308 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Using dynamic time warping distances as features for improved time series classification (2016, methodology): “Under the feature-based methods, a statistical (Nanopoulos et al. 2001; Geurts 2001) or symbolic (Lin et al. 2007; Ye and Keogh 2009) feature representation is first defined for time series.”
Characteristic-Based Clustering for Time Series Data (2006, methodology): “Nanopoulos and others (Nanopoulos et al., 2001) also suggested four different statistical time series measures for firstorder and second-order decompositions; thus, a total of eight features were used to form the input vector for a neural network classifier for a control chart classification applica”
Time Series Classification Using Multi-Channels Deep Convolutional Neural Networks (2014, methodology): “Some feature-based models have been explored for time series classification [2, 23], however, most of previous works extracted the hand-crafted statistical features based on domain knowledge, and achieved the performance not as well as sequence distance based models.”
Rotation-invariant similarity in time series using bag-of-patterns representation (2012, methodology): “Nanopoulos et al. (2001) proposed extracting statistical features of time series such as skewness, mean, variance, and kurtosis, and classifying the data using multi-layer perceptron (MLP) neural network.”
A review of unsupervised feature learning and deep learning for time-series modeling (2014, background): “The use of feature extraction has a number of advantages [97].”
Exploiting multi-channels deep convolutional neural networks for multivariate time series classification (2015, methodology): “Though many feature-based models have been explored for time series classification [6, 38], most of previous work extracted the hand-crafted statistical features based on domain knowledge, and achieved the performance not as well as sequence distance based models.”
Highly Comparative Feature-Based Time-Series Classification (2014, background): “As with the choice of representations and distance metrics for time series, features for time-series classiﬁcation problems are usually selected manually by a researcher for a given dataset.”
Forecasting: theory and practice (2020, background): “(Nanopoulos et al., 2001; Fulcher et al., 2013), anomaly detection (Kang, 2012; Talagala et al.”
catch22: CAnonical Time-series CHaracteristics (2019, background): “From among this wide range of possible features, selecting a set of features that successfully captures the dynamics relevant to the problemat handhas typically been donemanually without quantitative comparison across a variety of potential candidates (Timmer et al. 1993; Nanopoulos et al. 2001; Mör”
A Bag-of-Features Framework to Classify Time Series (2013, methodology): “[39] proposed a multi-layer perceptron neural network fed by statistical features such as means and standard deviations calculated from the time series.”
Fast time series classification using numerosity reduction (2006, methodology): “Many algorithms have been proposed for time series classification, including decision trees (Rodriguez & Alonso, 2004), neural networks (Nanopoulos & Manolopoulos, 2001), Bayesian classifiers, SVM (Wu & Chang, 2004), etc.”
A time series forest for classification and feature extraction (2013, methodology): “[11] extracted statistical features such as the mean and deviation of an entire time series, and then used a multi-layer perceptron neural network for classification.”
Visualising forecasting algorithm performance using time series instance spaces (2017, methodology): “The idea of characterising a time series as a feature vector is not new, and has been used for classifying time series (Nanopoulos, Alcock, and Manolopoulos, 2001), clustering time series (Wang, Smith, and Hyndman, 2006), and for identifying outlying or anomalous time series (Hyndman, Wang, and…”
LSTM Fully Convolutional Networks for Time Series Classification (2017, methodology): “Several approaches, including feature-based [7], ensembles [8]–[10], and deep learning [11], [12], have been utilized to classify time series.”