Abstract
When dealing with a new time series classification problem, modellers do not know in advance which features could enable the best classification performance. We propose an evolutionary algorithm based on grammatical evolution to attain a datadriven featurebased representation of time series with minimal human intervention. The proposed algorithm can select both the features to extract and the subsequences from which to extract them. These choices not only impact classification performance but also allow understanding of the problem at hand. The algorithm is tested on 30 problems outperforming several benchmarks. Finally, in a case study related to subject authentication, we show how features learned for a given subject are able to generalise to subjects unseen during the extraction phase.
Similar content being viewed by others
1 Introduction
Feature extraction should normally be customised to the problem at hand. Often, when modellers deal with a new problem they do not know which features could enable the best classification performance. Thus, a common approach is to construct an initial set of features, and then select the subset yielding best performance [34].
In contrast to the manual approach, there is growing interest in algorithms that enable the datadriven discovery of features, as made possible by deep learning methods [28]. The advantage is that modellers can redirect their efforts from the construction of the solution to the construction of the learning framework. While the former may be useful solely on a particular problem the latter may be effective on many.
In the present study, we investigate grammatical evolution (GE) [43], an evolutionary algorithm related to genetic programming (GP) [2, 27], as a means to achieve datadriven feature extraction from time series in the context of classification.
The featurebased approach to time series classification (TSC) is convenient for a variety of reasons [15, 39, 54]. (1) To reduce data to a manageable size. In particular, this can mitigate the curse of dimensionality, reduce computational requirements, and allow visualisation of time series datasets. (2) To highlight properties of a class of time series enabling understanding of the problem at hand. (3) To reduce the impact of noise and missing values. (4) To deal with time series of different length.
Our algorithm sequentially extracts a userdefined number of features. At each step, a run of the algorithm returns a feature that is intended to minimise the classification error while being minimally correlated with features extracted during previous steps. By combining a set of primitive functions e.g. mean/standard deviation, GE creates more complex functions we refer to as featureextractors. Each featureextractor takes a time series as input and returns a single feature (a scalar). This can be seen as combining feature extraction and feature selection within the same algorithm.
Previous research on TSC proved that class membership may depend on features related to the whole time series, or on features related to one or more of its subsequences [5]. Thus, we enable our algorithm to search for both the features to extract and the subsequences from which to extract them. We show that these choices are central not only for classification performance, but also to allow understanding.
The classification performance of extracted features is evaluated using a oneclass classifier [37] (Sect. 3.4). Oneclass classification is concerned with learning a classifier when all training samples belong to a single class. We are motivated by the requirements of a subject authentication problem we have recently investigated [36]. The aim of subject authentication is to confirm the identity of a person. Both binary and multiclass methods assume, in a sense, a fixed population of subjects wellrepresented in the data, which is not realistic for this scenario. Thus, the best option may be a semisupervised oneclass classifier tailored to the intended subject only.
Although we use a oneclass classifier, our algorithm requires labelled data to evaluate the fitness function that drives the extraction process. This is different from a pure oneclass classification scenario where only the samples of a single class are available for training a classifier. However, we use our subject authentication problem to show how features evolved using a oneclass classifier are able to generalise to classes unseen during the extraction phase (Sect. 6.1).
We compare our algorithm against a 1nearest neighbour classifier equipped with dynamic time warping (1NNDTW) on raw data, considered as the standard benchmark in the literature [5] (Sect. 4.2). Finally, we conduct an experimental analysis to demonstrate the impact of interval/function selection on performance (Sect. 4.3).
The remainder of this work is organised as follows. In Sect. 2, we provide a brief introduction to featurebased TSC. In Sect. 3, we describe our GEbased algorithm. In Sect. 4, we describe the benchmark methods. In Sect. 5, we describe the experimental design. In Sects. 67, we analyse our results. Finally, in Sect. 8, we draw our conclusions, and discuss future work.
2 Related work
We briefly introduce feature extraction in Sect. 2.1. Works related to featurebased TSC are reviewed in Sect. 2.2, evolutionary approaches in Sect. 2.3.
The literature related to featurebased TSC reveals that there are only a few works that investigate the oneclass assumption. One of the major contribution of the present study is to expand the discussion on oneclass TSC.
The literature related to evolutionary approaches to TSC reveals a number of gaps that we seek to address. There is a lack of consistent comparison of proposed methods with relevant benchmarks. We address this issue by evaluating our approach on 30 problems, all but one from the UCR/UEA archive [10]. The UCR/UEA archive is the reference archive of problems for TSC researchers. Also, we compare our results against a 1NNDTW on raw data considered as the standard benchmark in the literature [5]. It appears that it is not clear how to evolve multiple features that are not redundant. We tackle this problem by sequentially extracting the features, and requiring that their individual classification performance is maximised, while their average correlation with previously extracted features is minimised. Finally, we analyse evolved solutions to expand the discussion on problem/solution interpretability enabled by evolutionary techniques (Sect. 7).
2.1 Overview
Features are distinctive aspects of something. In machine learning, feature extraction aims at finding the most informative set of features for a certain task [22]. However, in most cases, the features to use are not known in advance. Thus, modellers go through time consuming trial/error procedures to extract and select most informative features.
A featurebased representation affects the choice and performance of the classifier and vice versa [14]. For instance, the quality of a featurebased representation may be overestimated if overfitting occurs, however, it may be underestimated if the classifier relies on assumptions violated by the featurebased representation (e.g. a normality assumption for a Gaussianmixture classifier). In practice, expert practitioners first exploit their domain knowledge to extract a set of features. Then, they carry out a statistical analysis of the features before selecting an appropriate classifier. In contrast, our algorithm extracts features that are not only suited for the problem at hand, but are appropriate for the classifier used in the extraction process.
Finally, our algorithm allows modellers to extract features from time series without requiring a previous specific knowledge of the field. This is in line with an increasingly popular research area known as automated machine learning (AutoML) [19]. The goal of AutoML is to automate the typical machine learning pipeline (e.g. data preprocessing, feature extraction, model and hyperparameter selection) reducing the amount of experts’ work required to deploy a model.
2.2 Featurebased TSC

Description: Time series are transformed into featurevectors which are used to complete the classification task through any offtheshelf classifier.

Pros: Most featurebased approaches facilitate problem and solution interpretability, and ease computational complexity.

Cons: It is not trivial to identify distinctive features especially when a new problem is addressed. Most featurebased approaches are expected to be weak on problems where features can be shifted along the time axis.
There are only few works investigating oneclass TSC [37]. In a previous experimental study, we evaluate several approaches to derive dissimilaritybased representations for oneclass TSC [37]. In dissimilaritybased representations, given a time series, each feature corresponds to its dissimilarity (distance) from one of a set of “reference” time series. Results show that the choice of the pair dissimilarity measure and set of “reference” time series is key to classification performance.
Two key questions that arise when dealing with featurebased TSC concern the features to extract, and the subsequence from which to extract them. Neither answer is obvious. The feature extraction process can be manual [35], or automated [36]. Features can be as general as statistical moments, or more carefully designed for time series like the timereversal asymmetry statistic [49].
Assuming that the most effective features are not known in advance, some authors propose to extract several features and then retain only the best ones. Deng et al. [12] propose a treeensemble classifier called “time series forest” (similar to a random forest). In this case tree nodes calculate simple statistics on randomly selected subsequences. Building on this idea, Shifaz et al. [50] propose to exploit the strengths of successful TSC algorithms creating three novel splitting criteria. Their ensemble of decision trees achieves competitive performance. How to deploy their approach to the oneclass classification domain is an interesting topic for future work.
Dempster et al. [11] apply thousands of random convolutional filters on time series. Resulting feature maps are summarised through two adhoc statistics, then concatenated in vectors, and used for classification. This approach exploits the success of convolutional neural networks for TSC, as reported in a recent comprehensive study on deep learning methods for TSC [16]. Finally, the method relies on the implicit feature selection enabled by a ridge regression classifier in order to avoid overfitting.
Lubba et al. [34] consider 4791 features for TSC. Of these, they retain only 22 features because of their classification performance on considered datasets, and their minimal redundancy. This small subset named “catch22” represents the state of the art for featurebased TSC, and we use it to benchmark our algorithm.
Concluding, other approaches to featurebased TSC include dictionarybased methods, and graph features. In dictionarybased methods, also known as bagofpatterns methods [32], time series are transformed into strings using SAX [33]. Then, each feature corresponds to the frequency of occurrence of a specific subsequence (or word) within a time series. Graphbased techniques first transform a time series into a graph, then features are extracted from this new representation [30].
2.3 Evolutionary approaches
The majority of prior research on evolutionary techniques for TSC has applied to ECG time series [18], and sensor time series for fault detection [31]. Some studies, discussed below, propose general purpose feature extraction algorithms [15, 23, 52].
Eads et al. [15] use GE [43] to evolve a single population of featureextractors using a set of 25 primitives. Each featureextractor is able to target any subsequence and return a single scalar. This hillclimbing algorithm does not make use of any crossover operator. However, it allows modifications to the current solution (addition, deletion, mutation of featureextractors) only if changes increase the performance, or cause a negligible impact on performance but a decrease in runtime. Classification performance, tested on seven datasets, is better than that of raw data.
Harvey and Todd [23] propose an algorithm based on GP [27] where 35 primitives are used to evolve sets of featureextractors. Nearly all the primitives take a time series as input and output a transformed time series. To reduce time series to a single scalar each featureextractor must end with a summation. The authors provide a number of rules to reduce redundancy of final featureextractors and increase their interpretability. Performance is tested on simulated data only.
Finally, Virgolin et al. [52] propose a GPbased algorithm for sequential feature construction but not on time series. While feature extraction concerns the extraction of features from raw data, feature construction concerns the transformation of existing features. They focus on the interpretability of the GP trees emphasising interpretability of the original features, and limiting the height of the trees. As discussed in Sect. 7, we are also interested in interpretability. We point out that in TSC interpretability is aided by knowing which functions, and subsequences allow good classification performance.
3 Proposed method
In this section we describe our evolutionary algorithm for feature extraction from time series based on GE. The core components of our approach are the grammar (Sect. 3.2), and the fitness function (Sect. 3.3).
3.1 Overview
Our algorithm relies on GE, a grammarbased form of GP. We choose GE to implement our algorithm because, as opposed to GP, it allows us to handle a mixture of data types. While the type constraint could be handled with other approaches e.g. strongly typed GP [38], we believe that the grammar is a particularly convenient way to express the syntax of admissible solutions, and also we want solutions to be readable Python code as this can facilitate understanding.
The grammar allows the modeller to exploit her/his domain knowledge, and to impose syntactical constrains to guide the search of feasible solutions. In this study we have used our knowledge of the TSC domain to define the proposed grammar. We have designed the grammar to be balanced rather than explosive [41], giving a bias towards short solutions. In many cases, the primitives included in the grammar are highlevel functions specifically defined for time series e.g. the complexity estimate [6]. However, a practitioner specialising in a particular subdomain of TSC could add further domain knowledge, e.g. spectral features known to be useful in that domain. In addition, the grammar allows the selection of any subsequence. Subsequences may be key to class membership as demonstrated by TSC algorithms like shapelets [55], or bagofpatterns [32].
The solution space of our algorithm consists of featureextractors. Each featureextractor F is a function which takes as input a specific subsequence [a : b] of a time series T e.g. T[20 : 50]. The output is a single scalar i.e. a feature. In Eq. 1 is shown an example of a featureextractor. Note that all the primitives included in a given featureextractor are applied to the same subsequence.
Our algorithm outputs a single featureextractor. Thus, to extract multiple features we have to run the algorithm multiple times. Sequential extraction allows control over the “quality” of all features. If we were to evolve a number of features all at once we would not know which ones are contributing to classification performance, or which ones are even harming it. Furthermore, by extracting one feature at a time we are oriented towards finding the minimum number of features for the problem at hand. In fact, we can extract one feature at a time and stop as soon as we meet the required level of performance, or we do not observe any meaningful improvement. On the other hand, our greedy approach to feature extraction may not find the global optimum. The alternative to evolve a set of featureextractors at once, already considered by Eads et al. [15], warrants further investigation.
Concluding, the fitness function requires each feature to have good classification performance, while being minimally correlated with previous features. While the first component of the fitness function is defined to extract features of good predictive power, the second one weakens the redundancy of extracted features.
3.2 Grammar and primitives
The grammar guiding the search of featureextractors is shown in Fig. 1. Details of the encoding of candidate solutions, and the mapping process can be found in previous works [43].
The general idea is to compose a featureextractor \(\texttt {<F>}\) by selecting its individual components one at the time. Each component, enclosed in the angle brackets (\(\texttt {<.>}\)), requires a choice from a set of predefined equally probable alternatives, separated by pipe symbols (\(\texttt {}\)). Finally, the actual extraction is carried out by a wrapper function Extract that simply takes the input arguments and arranges them in the form shown in Eq. 1, returning a feature. Note that T (a time series) is not enclosed in the angle brackets. This is because T is an argument that we pass to the function at call time.
As said before, a featureextractor consists of a function \(\texttt {<fun>}\) applied to a subsequence of a time series T. To select a specific subsequence we need to choose a lower bound \(\texttt {<lb>}\) and an upper bound \(\texttt {<ub>}\). These indices are chosen from the range [1, L], where L corresponds to the time series length. Since a requirement is that the lower bound is below the upper bound, if this condition is violated \(\texttt {<lb>}\) and \(\texttt {<ub>}\) are swapped. The grammar allows \(\texttt {<ub>}\) to have a None value. In this case if \(\texttt {<bool>}\) is True we select the subsequence \([1:\texttt {<lb>}]\), if False the subsequence \([\texttt {<lb>}:L]\).
A function \(\texttt {<fun>}\) can use one or more elements of the \(\texttt {<primitive>}\) set. Multiple primitives are connected through the operators {\(+\), −, \(*\), AQ} contained in \(\texttt {<op>}\). In \(\texttt {<op>}\), AQ (analytic quotient) is a function designed to perform division while avoiding division by 0 error [40]. This is achieved by transforming a divisor d into \(\sqrt{a + d^2}\). We set \(a=1\) as suggested by the authors of this function.
In \(\texttt {<fun>}\), \(\texttt {<primitive>}\) is repeated twice while \(\texttt {<op>}\) only once. This is done to make the selection of \(\texttt {<primitive>}\) twice as likely than \(\texttt {<op>}\) thus interrupting the recursion triggered by the \(\texttt {<op>}\) rule.
The set of 17 primitives includes functions able to take a subsequence as input and output a single scalar. These are among the simplest and most commonly used functions in featurebased time series classification: mean, standard deviation, median, skewness, kurtosis, max, min. Above0 and Below0 count the number of values above/below 0. AbsoluteEnergy returns the sum of the absolute value. AR returns the coefficient of an autoregressive model of order 1. Autocorrelation returns the autocorrelation at lag 1. CE measures the shape complexity of a time series [6]. FFT returns the amplitude associated with the largest coefficient of a fast Fourier transform. LinearTrend returns the slope of a linear regression model. MeanChanges returns the mean of first order differences. TRAS is a measure of nonlinearity known as time reversal asymmetric statistic [49].
3.3 Fitness evaluation
The fitness function, shown in Eq. 2, drives the search through the solution space. The fitness score is crucial because it influences the probability of a featureextractor to be selected, breed with other featureextractors, and thus pass on to future generations. The fitness evaluation process is illustrated in Fig. 2. Our goal is to sequentially extract features that are all able to achieve a good classification performance individually, while having minimum linear dependence between them.
As an example, suppose we want to extract two features for a certain classification problem. This means we have to run our algorithm \(x=2\) times with each run outputting a single featureextractor. The first featureextractor is evolved aiming at the minimisation of the classification error on a validation set. The classification error is defined as \(1\)AUROC. On the other hand, the second featureextractor is evolved aiming for both minimisation of the classification error on a validation set, and minimisation of the average squared Pearson correlation coefficient with previous features \(\overline{R^{2}}(F)\). This coefficient is calculated as follows. We take the featurebased representation of training data according to a candidate featureextractor. Then, we calculate its squared Pearson correlation with the featurebased representation of the same data according to each of the featureextractors outputted from previous runs (\({\hat{F}}_{i}\)). Finally, we calculate the average value. Both the classification error, and the average squared Pearson correlation coefficient have magnitude in the range [0, 1] thus they are commensurate in scale, and have the same impact on the fitness score. This process is repeated three times using 3folds crossvalidation. Thus, the final fitness score of a featureextractor corresponds to the average score of the 3folds.
In summary, after all the featureextractors of an initial population are assigned a fitness score, the crossover and the mutation operators are applied to create a new generation. The evolutionary process continues until a certain number of generations is reached. At that point the featureextractor with the lowest fitness is considered the final solution of the algorithm.
In Eq. 4\(cov(\cdot )\) and \(\sigma (\cdot )\) represent the covariance and the standard deviation functions respectively.
3.4 Classification framework
Oneclass classification [26], closely related to anomaly detection [8], is concerned with learning a classifier when only the data of a single class is available at training time, and/or at prediction time we might be exposed to classes that are unknown at training time. Usually, the class we learn is referred to as the positive (normal) class while all the samples that fall outside it are allocated to a “generic” negative (anomalous) class. Figure 3 is designed to demonstrate the usefulness of the oneclass assumption. In part (1), a binary classifier trained to separate positive/negative samples fails when a new class of negative samples appears at prediction time. The same reasoning can be extended to a multiclass classifier. Conversely, in part (2) a oneclass classifier trained on positive samples only is robust to negative classes whether they are known or not at training time.
There are several real world applications where data from the negative class are difficult or impossible to obtain. For instance, in biometric subject authentication problems [36] it is relatively easy to collect some data from the “target subject” class, however it is less straightforward to collect data that are representative of the “not target subject” class. Other related applications are: fraud detection [44], machine fault detection [51], medical diagnosis [48], network intrusion [7], and signature verification [21]. This problem is little investigated in the context of TSC [37].
The oneclass classifier we use both in the feature extraction phase, and to evaluate the quality of extracted features is the 1NN classifier equipped with Euclidean distance (1NNED). The idea is to use a simple nonparametric classifier to emphasise the quality of extracted features, and make minimal assumptions about class distributions. During the prediction phase, a sample is assigned a score equal to the distance from its nearest training sample. By imposing a threshold on the distance, a sample can be classified as either positive or negative. A common way to set the threshold is to choose a value such that 95% of training data is correctly labelled as below the threshold, hence positive. However, this approach can require some fine tuning according to the specific problem at hand.
In order to avoid this and have a method that can be consistently applied across all datasets we use classification scores to calculate the AUROC. The AUROC is obtained by computing the underlying area of a curve constructed by plotting the true positive rate against the false positive rate at various threshold settings. Threshold values are calculated as the midpoint between each pair of sorted classification scores.
We believe the AUROC is particularly suitable for oneclass TSC experiments since it is insensitive to class imbalance. Furthermore, Gay and Lemaire [20] advocate the use of the AUROC for binary and multiclass TSC experiments too.
4 Benchmark methods
In this section we describe the benchmark methods we compare with. We consider a random search (Sect. 4.1), a distancebased approach (Sect. 4.2), and a featurebased approach with some variants (Sect. 4.3).
4.1 Random search
Random search (RS) is often used to evaluate whether the evolutionary process brings any benefit over the random generation of solutions. Thus, it disentangles the effect of the grammar from the effect of the search process.
Following the rules of our grammar, the RS creates a number of featureextractors (# generations \(\times \) population size, Sect. 5.2) with no duplicates allowed. Featureextractors are evaluated through our fitness function, and the best one is selected for comparison with our algorithm. As per our algorithm, the RS is repeated multiple times in order to extract multiple features.
4.2 1NN with DTW
The 1NNDTW classifier (with warping window set through crossvalidation [47]) is considered the standard benchmark in the TSC literature [5]. Unlike all the other approaches used in this study 1NNDTW does not involve featurevectors but raw time series. This is a distancebased classification approach that compares two time series using DTW. In the oneclass classification framework each test sample receives a classification score equal to the DTW distance to its nearest training sample.
For the warping window size we have tested all the values in the range \([1,0.1{\times }L]\) (where L is the time series length) using a 10fold crossvalidation on validation data.
4.3 Analysis of function/subsequence selection
As discussed in Sect. 3, our algorithm can dynamically decide both the features to extract and the subsequences from which to extract them. A parallel can be drawn with the manual approach where features and subsequences can be selected according to fixed userdefined strategies. To investigate the behaviour of our datadriven algorithm we evaluate a Cartesian product of the alternatives mentioned below.
In terms of features to be extracted we consider two sets. (1) The 17 primitives used in our algorithm (Sect. 3.2) and (2) the C22 features discussed in Sect. 2.2. Overall, C22 features are different and more complex with respect to our primitives.
In terms of segmentation strategies we consider three approaches. (1) We extract the features from the whole time series. (2) We segment time series into a number of (approximately) equally sized subsequences. Then, features are extracted from each subsequence and concatenated into a featurevector. (3) We segment time series using a change point model. Generally speaking, change point models are used to divide a time series into distinct homogeneous subsequences [4]. We use the bottomup segmentation algorithm proposed by Keogh et al. [25]. In order to define the number of subsequences and their boundaries for a given dataset we implement the following algorithm. Given a training set we select 50% of the samples at random. We segment each sample using the bottomup algorithm and we record the number of subsequences found. Then, we compute the average number of subsequences (rounding to the nearest integer). Continuing, we rerun the segmentation algorithm on the same samples imposing as stopping criterion the average number of subsequences found before. Finally, we average resulting indices.
5 Experimental design
We provide an overview of the datasets used in Sect. 5.1. A summary of the main components of the evolutionary framework is provided in Sect. 5.2. We report some implementation details in Sect. 5.3.
5.1 Time series data
We test our algorithm on 30 datasets. Of these, 29 are selected from the UCR/UEA archive, while one is a proprietary dataset [36]. All the datasets are partitioned into labelled training and test sets and have previously been examined in several binary and multiclass TSC experiments [5]. All the time series are univariate, contain only real numbers, have a fixed length within a given dataset, and are znormalised [46].
We adapt the datasets to the oneclass classification framework as follows. Given a dataset, while maintaining the original split between training/test data, each of the classes is considered in turn as the positive class (and so it is used for training) and all the others become the negative class. Classification performance for a given dataset corresponds to the average performance over the classes.
When carrying out a TSC experiment it is good practice to consider all the datasets of the UCR/UEA archive [10]. This is to avoid spurious results, and allow a thorough comparison between different studies. However, we consider only 29 datasets selected as follows: from the original body of 85 datasets of the archive, first we filter out those where there is at least one class with less than 17 training samples. Then we sort the remaining ones according to their total number of samples and select the smallest 29. The rationale is that our algorithm requires a validation set to evaluate the quality of candidate featureextractors during the evolutionary process. Thus, we want to ensure that there is always a sufficient amount of samples to allow a meaningful training/validation split. Specifically, we use a 2:1 split of training data. Furthermore, we prefer smaller datasets as our current implementation is very expensive in terms of runtime.
5.2 GE configuration
The evolutionary process is configured with parameters that are common across the field [45]. The algorithm follows a generational approach with a population of 500 individuals (featureextractors) for 40 generations. In the GP literature it is most common to use 50 generations [27], however 40 is also often used [9]. In their review of the GP literature, Poli et al. [45] state that the number of generations typically falls in the range [10, 50], where the most productive search is usually performed.
The population is randomly initialised with uniform probability. Genomes are initialised with length 200, while the max length is set to 500. Individuals are recombined with each other using twopoint crossover with probability 0.8, and no wraps are allowed. Given two individuals, the crossover operator creates two new individuals by randomly selecting two different points on each genome. Then, the initial and final sections of one genome are merged with the mid section of the other. We increase the selection pressure setting the tournament size to five. After the crossover phase, the resulting population is subject to the mutation operator. We use the intflip mutation operator that may change with probability 0.01 each gene of a given individual. Elitism is used to preserve the best individual through the generations.
5.3 Implementation details
The experiment is implemented in the Python programming language^{Footnote 1} and relies on the PonyGE2 library [17].
The experiment took about 5000 CPU hours. Our current implementation is expensive especially if compared with the benchmark methods which take just a few hours to run. However, it is not straightforward to draw conclusions about the computational complexity, and scalability of our algorithm. First, for each dataset we run the algorithm 30 times for statistical purposes. However, this would not be required in realworld use. Continuing, runtime depends on several implementation details (e.g. the primitives included in the grammar) which could be reconsidered in future research. Finally, the expensive feature extraction phase is balanced by savings during the prediction phase. In fact, during the prediction phase our algorithm allows the classifier to work with lowdimensional featurebased representations. In contrast, 1NNDTW always requires the whole time series to work.
6 Results
Classification performance for the datasets of the UCR/UEA archive is shown in Tables 1, 2. Results related to our subject authentication problem are discussed in Sect. 6.1. For each dataset we extract 10 features and we repeat this process for 30 independent runs. We standardise features to have zero mean and unit variance (with reference to the training set). Algorithm limitations are discussed in Sect. 6.2.
We compare the performance of the features evolved through our algorithm with that of two benchmark sets of features. Of these, PR is the set of primitives used in the grammar, while C22 is the set of features selected by Lubba et al. [34]. These features are used to derive featurebased representations of time series according to three different strategies. (1) Features are extracted from the whole time series (columns 1PR/1C22). (2) Time series are broken down into 5 adjacent and equally sized subsequences. Then, features are extracted from each subsequence and grouped together in a featurevector (columns 5ESPR/5ESC22). (3) Time series are broken down in a number of adjacent subsequences using a change point model [25]. Then, features are extracted from each subsequence and grouped together in a featurevector (columns CPMPR/CPMC22). We observe that the average number of subsequences per dataset found through the change point model is 5 and this is why we use this number in item (2). Concluding, we also compare against a distancebased classification approach (column 1NNDTW), a strong baseline [5].
Results show that on average our algorithm is superior to any benchmark method. Our algorithm gives better results than 1NNDTW. It also improves on the C22 features considered the state of the art for featurebased TSC [34].
We consider column 13 our best result (in the tradeoff between best average performance subject to the lowest dimensionality achievable, which is our aim), while 1NNDTW is the best benchmark method for comparison. We find that the difference between column 13 and 1NNDTW is statistically significant using a Wilcoxon signedrank test with a pvalue threshold of \(\alpha =0.05\) [53]. We observe that the pvalues decrease monotonically as additional features are added, as we would expect. If a BonferroniHolm correction for multiple testing [1] is applied, statistical significance is achieved for all tests from 12 onward. As 13 is the point with the largest reduction in pvalue relative to the preceding feature number choice (13, pvalue\(=\)0.007 vs. 12, pvalue\(=\)0.024), we select this as our optimal result in the performance/dimensionality reduction tradeoff .
Comparing the average performance of columns 1–3 and 1PR in Table 1, we can see that our algorithm is able to exploit and combine the functions included in the set of primitives in a way that allows \(+17\%\) AUROC relative to the use of the functions alone. Furthermore, our algorithm achieves the highest average performance using only 3 features. This gives the time series representation with lowest dimensionality. In fact, PR and C22 require at least 17 and 22 features respectively while 1NNDTW requires the whole time series. Dimensionality has an impact on storage requirements, and computational complexity at prediction time.
To extract features from a segmented time series has a positive impact on performance. We note a \(+11\)/\(10\%\) AUROC comparing the average performance of column 1PR with that of columns 5ESPR/CPMPR. Conversely, the same does not hold for C22 features. In that case segmentation has little or no impact on performance. This seems to depend on the nature of features. The PR set mainly includes simple statistical moments like mean, standard deviation etc. Although these features are easy to compute, their descriptive power is weak when they are extracted from the whole time series. Moreover, all time series of the UCR/UEA archive are znormalised so they all have 0 mean and a standard deviation of 1. When a time series is segmented, PR features have more descriptive power.
On the other hand, C22 features are more complex, thus they are able to capture some distinctive characteristics even when they are extracted from the whole time series. In order to evaluate if 1C22 features are overfitting we measure their performance on validation data. The average performance across all datasets on validation data is equal to 66% AUROC. Considering that the performance of 1C22 features on test data is 70% AUROC we can conclude that these features are not overfitting.
On average, the performance achieved by segmenting a time series in equally sized subsequences is nearly the same as that achieved by segmenting through a more complex model, as we can see comparing columns 5ESPR/5ESC22 with columns CPMPR/CPMC22. It is difficult to explain such a result within this experimental study. Investigating this result in more detail is the focus of further research. However segmentation in equally sized segments can be considered as a strong baseline.
Our method converges to the maximum performance with only three features. In Fig. 4 we show the average AUROC for all datasets of our algorithm (filled circle). As shown in part (1), the “quality” of features decreases by \(4\%\) from the first to the third feature. The performance remains steady from the third to the tenth feature (when rounded to the nearest integer). This demonstrates that individually features maintain a similar level of performance. In part (2), we can see that by combining multiple features the AUROC increases by \(2\%\) between the first and the third feature. Then we can observe a weak but positive trend. Features extracted by our algorithm are not explicitly required to improve the overall classification performance when grouped with those that have been extracted before. This objective is pursued indirectly by minimizing the linear correlation of a feature with those extracted before.
Secondly, Fig. 4 shows RS average AUROC for all datasets (\(\blacksquare \)). In order to speedup the computation we do not repeat the RS for 30 runs, as done for our algorithm. The variance between RS runs is expected to be low because the sample size is large, and there is no danger of an “unlucky” initial population, as in GE itself. Also, as we have a large enough number of datasets, the variance across datasets makes our results statistically significant. Our algorithm is better than RS. However, RS performance rivals 1NNDTW, a strong baseline. This demonstrates that both our grammar and the search contribute strongly to performance.
6.1 Subject authentication
This proprietary dataset, which we refer to as “AccelerometerData”, concerns a subject authentication problem [36]. To collect this data a group of nine subjects wore a watchlike triaxial accelerometer for about a month in free living conditions. Each time series corresponds to the average acceleration per minute over an entire day.
Features extracted by our algorithm enable good classification performance when tested on data that include subjects not available during the feature extraction phase. In other words, features learned to distinguish subject 1 from subjects 26 are also effective to distinguish subject 1 from subjects 79.
Figure 5 shows the average classification performance achieved by our algorithm on a subset of subjects (16). The solid line (solid line) represents the performance achieved when, during the featureextraction phase, we use a validation set that includes only subjects 16. The dashed line (dashed line) represents the performance achieved when, during the featureextraction phase, we use a validation set that includes all subjects (1–9). In both cases, extracted features are evaluated on a test set including all subjects (1–9). The two lines follow almost the same trajectory. Excluding the first feature, the average absolute difference between the two lines is equal to 0.3% AUROC.
6.2 Limitations
Our algorithm reveals a number of limitations. The first limitation concerns the runtime required by our current implementation. Although, as discussed in Sect. 5.3, there are several aspects to consider in this regard, overall the algorithm would greatly benefit from any substantial improvement of its runtime.
Continuing, we observe that the algorithm tends to overfit. As shown in Fig. 6 part (1), average AUROC of features 1 to 10 on validation data is approximately 11% higher than on test data. This value drops to 6% if we consider sequences of features, as shown in part (2). The performance per feature seems to follow the same pattern on both validation and test data. Conversely, the performance per group of features exhibits a weak negative trend on validation data, and a weak positive trend on test data showing that together multiple features generalise better. Also, the performance of our algorithm remains better than baselines on test data. While overfitting is a problem it also presents an opportunity to improve our results through regularisation strategies, as we plan to do in future work.
Finally, one possible weakness in our fitness function is that as the number of extracted features increases the correlation penalty is averaged over an increasing number of features as well. While this could make the penalty very small, after a large number of featureextractors have been created, we observe that a small number of features (e.g. 3 in Table 1) achieves best average performance.
7 Featureextractors and interpretability
In this section we show how the evolutionary process leads to the selection of specific features and subsequences according to the problem at hand. This not only enables good classification performance but also understanding.
The development of interpretable machine learning models is an important research topic [13]. Interpretability is not only essential to machine learning practitioners, e.g. understanding a model allows understanding of its limitations, but also legislation can require model interpretability. For instance, the General Data Protection Regulation act^{Footnote 2} introduced in the European Union gives to citizens a “right to explanation” with respect to decisions taken by algorithms using their data.
Several researchers have shown that GP can be used to enable interpretability of socalled blackbox models (models that are not easy to interpret) [52]. Others have shown that evolutionary algorithms can be as effective as blackbox models with the advantage of being more interpretable [29].
Our algorithm allows interpretability of the classification outcome in several ways. In Sects. 7.1, 7.2, we expand the discussion about featureextractors and interpretability in the context of TSC, however here we provide some more general considerations. First, as the search for featureextractors is driven by the classification performance they allow through a 1NN classifier, it is expected that our approach considers a sample to be normal if it is “close enough” to other samples we already know to be normal (i.e. training data). This type of decision by analogy resembles human thinking [42]. Continuing, extracted features are immediately interpretable as they are expressed as readable Python code. Also, as extracted features are composed of highlevel functions e.g. mean, standard deviation, etc. predictions made by our algorithm can be explained in the context of the problem at hand. Finally, as our algorithm gives good classification performance with two or three features, it allows useful visualisation of time series datasets (Figure 9), and data visualisation is considered key to interpretability [29]. On the other hand, as the number of primitives included in a featureextractor increases they become more difficult to interpret (Appendix A). As suggested by Lensen et al. [29] in the context of data visualisation, future work may investigate multiobjective fitness functions able to tradeoff between classification performance and complexity of the solutions found.
7.1 The features to extract
We use the “SyntheticControl” dataset [3] as a case study to illustrate the ability of our algorithm to discover the right features for the problem. As shown in Fig. 7 this dataset contains 6 different classes of simulated time series. In (1) time series are generated by sampling from a Normal distribution. In (2) time series are generated by sampling from a Normal distribution and adding a cyclic component through a sine function. In (3) and (4) time series are generated by sampling from a Normal distribution and adding/subtracting a trend component. In (5) and (6) time series are generated by sampling from a Normal distribution and adding/subtracting a trend component only after/before a certain time point.
We count the selection frequency of each function/operator used in the first two featureextractors evolved for each class of the “SyntheticControl” dataset. Without loss of generality we focus on the first two featureextractors because this eases our discussion and enables visualisation as shown in Fig. 9. Featureextractors used to generate Fig. 9 are detailed in Appendix A.
In Fig. 8 we can see that the algorithm selects specific groups of functions/operators for each class of the dataset. As reflected by the bar heights, the algorithm finds that class 1 is mainly characterised by the coefficient of an autoregressive model (AR), and a measure of complexity (CE). The cyclic component of class 2 is captured through the FFT and the autocorrelation functions. Classes 3 and 4 requires similar features like max/min, and the mean. In addition, class 4 is characterised by the linear trend. Finally, classes 5 and 6 require functions like linear trend, max/min, mean, and number of values above/below zero. Finally, the algorithm most often relates multiple functions through the AQ operator for all classes.
Concluding, as mentioned before, features are standardised (with reference to the training set) hence as shown in Fig. 9 our algorithm tends to concentrate the samples of the normal class around the origin even though this is not explicitly required. This aspect shows that our algorithm could also work well using “simple” classifiers, for instance a radial basis function classifier which has the advantage of requiring only a single hyperparameter, i.e. a distance threshold from the origin.
7.2 The subsequences from which to extract
To offer insight into which subsequences are most useful we consider the “GunPoint” dataset [55], and our “AccelerometerData” dataset [36] (Figs. 1012).
Figure 10 is derived by averaging over the two classes of the “GunPoint” dataset. In this case, our algorithm overselects two specific subsequences. The most frequently selected one corresponds to the subsequence underlying the large peak at the begin of the dotted line. The other corresponds to the subsequence underlying the small peak towards the end of the dotted line. Ye and Keogh [55] demonstrate that the time series of this dataset can be classified with high performance using “shapelets”. Shapelets are subsequences that are maximally representative of a class according to a predefined criterion [55]. In another study related to shapelets Hills et al. [24] show that there are two important subsequences in the “GunPoint” problem, the same are identified by our algorithm. Also, they report that the top five most important shapelets extracted by their algorithm are related to the end of the time series. However, our algorithm overselects the subsequences at the beginning of the time series. This means that our algorithm is able to target subsequences that are relevant for the problem, but overall it works in a way that is different from shapelets. Thus, the representation generated by our algorithm could be a good addition to a representation based on shapelettransform [24].
In Fig. 11, we show all the subsequences of the “GunPoint” dataset related to the time interval [0 : 40]. This is the most selected time interval according to the analysis related to Fig. 10. Subsequences are divided per class: part (1) and (2) show the subsequences related to class 1 and 2 respectively. The figure provides insights into the subsequences that are important for classification according to our algorithm. Visual assessment of the subsequences allows us to claim that most time series could be correctly classified if we were able to catch shape dissimilarities between the two classes in the considered time interval. As shown in Table 1, our algorithm achieves approximately 90% AUROC on this dataset. Thus, not only our algorithm is able to detect important subsequences, but also it is able to extract useful features.
Finally, Figure 12 considers all the nine classes of the “AccelerometerData” dataset. We can see that each person is characterised by her/his activity during a specific part of the day. In the time axis 0 corresponds to 0am and 1440 to 11.59pm. Most subjects are characterised by their activity between 6.30–8.30am, presumably the time they wake up and go to work. Two subjects are characterised by their activity during 0–6.30am (1,6). Also, for three subjects (6,7,9) the activity between 9–11pm seems to be important for their classification.
8 Conclusions
We propose a datadriven evolutionary algorithm for feature extraction from time series. The goal is to find a lowdimensional featurebased representation that enables good oneclass classification performance with minimum human intervention.
The search for suitable solutions is guided by our grammar specifically defined for time series. In fact, our algorithm can select both the features to extract, and the subsequences from which to extract them. Both aspects can be key to TSC problems, as revealed by several related studies.
Evolved features are composed of highlevel functions e.g. mean, standard deviation, etc. By choosing the number of features to be extracted a highdimensional time series can be reduced to a featurevector of arbitrary dimensionality. This not only lowers computational complexity at prediction time, but also allows the visualisation of time series datasets. Furthermore, the analysis of the functions/subsequences that are most frequently selected during the evolutionary process enables understanding about the problem at hand. Finally, our algorithm enables better classification performance than considered benchmark methods.
Future work could investigate further the generalization capabilities of our algorithm in a oneclass classification scenario where new classes appear at prediction time. Another avenue of research could investigate other primitives to be included in the grammar. A remaining issue of our algorithm is the tendency to overfit. This suggests that our results could be improved, an interesting topic for future work.
References
H. Abdi, Holm’s sequential bonferroni procedure. Encycl. Res. Des. 1(8), 1–8 (2010)
A. Agapitos, R. Loughran, M. Nicolau, S. Lucas, M. O’Neill, A. Brabazon, A survey of statistical machine learning elements in genetic programming. IEEE Trans. Evol. Comput. 23(6), 1029–1048 (2019)
R.J. Alcock, Y. Manolopoulos, et al., Timeseries similarity queries employing a featurebased approach. In: 7th Hellenic conference on informatics, pp 27–29 (1999)
S. Aminikhanghahi, D.J. Cook, A survey of methods for time series change point detection. Knowl. Inf. Syst. 51(2), 339–367 (2017)
A. Bagnall, J. Lines, A. Bostrom, J. Large, E. Keogh, The great time series classification bake off: a review and experimental evaluation of recent algorithmic advances. Data Min. Knowl. Discov. 31(3), 606–660 (2017)
G.E. Batista, E.J. Keogh, O.M. Tataw, V.M. De Souza, Cid: an efficient complexityinvariant distance for time series. Data Min. Knowl. Discov. 28(3), 634–669 (2014)
V.L. Cao, M. Nicolau, J. McDermott, Learning neural representations for network anomaly detection. IEEE Trans. Cybern. 99, 1–14 (2018)
V. Chandola, A. Banerjee, V. Kumar, Anomaly detection: a survey. ACM Comput. Surv. (CSUR) 41(3), 15 (2009)
W. Cui, A. Brabazon, M. O’Neill, Evolving efficient limit order strategy using grammatical evolution. In: IEEE Congress on Evolutionary Computation, IEEE, pp 1–6 (2010)
H.A. Dau, A. Bagnall, K. Kamgar, C.C.M. Yeh, Y. Zhu, S. Gharghabi, C.A. Ratanamahatana, E. Keogh, The ucr time series archive. IEEE/CAA J. Automatica Sinica 6(6), 1293–1305 (2019)
A. Dempster, F. Petitjean, G.I. Webb, Rocket: exceptionally fast and accurate time series classification using random convolutional kernels. Data Min. Knowl. Discov. 34(5), 1454–1495 (2020)
H. Deng, G. Runger, E. Tuv, M. Vladimir, A time series forest for classification and feature extraction. Inf. Sci. 239, 142–153 (2013)
F. DoshiVelez, B. Kim (2017) Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:170208608
R.O. Duda, P.E. Hart, D.G. Stork, Pattern Classification (Wiley, NewJersey, 2012)
D. Eads, K. Glocer, S. Perkins, J. Theiler, Grammarguided feature extraction for time series classification. In: Proceedings of the 9th Annual Conference on Neural Information Processing Systems, Citeseer, pp 1–8 (2005)
H.I. Fawaz, G. Forestier, J. Weber, L. Idoumghar, P.A. Muller, Deep learning for time series classification: a review. Data Min. Knowl. Discov. 33(4), 917–963 (2019)
M. Fenton, J. McDermott, D. Fagan, S. Forstenlechner, E. Hemberg, M. O’Neill, Ponyge2: Grammatical evolution in python. In: Proceedings of the Genetic and Evolutionary Computation Conference Companion, ACM, pp 1194–1201 (2017)
E. FernándezBlanco, D. Rivero, M. Gestal, J. Dorado, Classification of signals by means of genetic programming. Soft Comput. 17(10), 1929–1937 (2013)
M. Feurer, A. Klein, K. Eggensperger, J. Springenberg, M. Blum, F. Hutter, Efficient and robust automated machine learning. In: Advances in neural information processing systems, pp 2962–2970 (2015)
D. Gay, V. Lemaire, Should we Reload Time Series Classification Performance Evaluation? (a position paper). ArXiv:1903.03300 [stat.ML] (2019)
Y. Guerbai, Y. Chibani, B. Hadjadji, The effective use of the oneclass svm classifier for handwritten signature verification based on writerindependent parameters. Pattern Recognit. 48(1), 103–113 (2015)
I. Guyon, A. Elisseeff, An Introduction to Feature Extraction In Feature Extraction (Springer, Berlin, 2006)
D.Y. Harvey, M.D. Todd, Automated feature design for numeric sequence classification by genetic programming. IEEE Trans. Evol. Comput. 19(4), 474–489 (2015)
J. Hills, J. Lines, E. Baranauskas, J. Mapp, A. Bagnall, Classification of time series by shapelet transformation. Data Min. Knowl. Discov. 28(4), 851–881 (2014)
E. Keogh, S. Chu, D. Hart, M. Pazzani, Segmenting time series: A survey and novel approach. In: Data mining in time series databases, World Scientific, pp 1–21 (2004)
S.S. Khan, M.G. Madden, Oneclass classification: taxonomy of study and review of techniques. Knowl. Eng. Rev. 29(3), 345–374 (2014)
J.R. Koza, Genetic programming: on the programming of computers by means of natural selection, vol 1. MIT press (1992)
Y. LeCun, Y. Bengio, G. Hinton, Deep Learn. Nat. 521(7553), 436 (2015)
A. Lensen, B. Xue, M. Zhang, Genetic programming for evolving a front of interpretable models for data visualization. IEEE Transactions on Cybernetics (2020)
D. Li, J. Lin, TFDA. Bissyande, J. Klein, Y. Le Traon, Extracting statistical graph features for accurate and efficient time series classification. In: 21st International Conference on Extending Database Technology (2018)
L. Liao, Discovering prognostic features using genetic programming in remaining useful life prediction. IEEE Trans. Ind. Electron. 61(5), 2464–2472 (2014)
J. Lin, Y. Li, Finding structural similarity in time series data using bagofpatterns representation. In: International conference on scientific and statistical database management, Springer, pp 461–477 (2009)
J. Lin, E. Keogh, S. Lonardi, B. Chiu, A symbolic representation of time series, with implications for streaming algorithms. In: Proceedings of the 8th ACM SIGMOD workshop on Research issues in data mining and knowledge discovery, ACM, pp 2–11 (2003)
C.H. Lubba, S.S. Sethi, P. Knaute, S.R. Schultz, B.D. Fulcher, N.S. Jones, catch22: canonical timeseries characteristics. Data Min. Knowl. Discov. 33(6), 1821–1852 (2019)
S. Mauceri, L. Smith, J. Sweeney, J. McDermott, Subject recognition using wristworn triaxial accelerometer data. In: International Workshop on Machine Learning, Optimization, and Big Data, Springer, pp 574–585 (2017)
S. Mauceri, J. Sweeney, J. McDermott, Oneclass subject authentication using feature extraction by grammatical evolution on accelerometer data. In: Heuristics for Optimization and Learning, Springer, pp 393–407 (2018)
S. Mauceri, J. Sweeney, J. McDermott, Dissimilaritybased representations for oneclass classification on time series. Pattern Recognit. 100, 107122 (2020)
D.J. Montana, Strongly typed genetic programming. Evol. comput. 3(2), 199–230 (1995)
A. Nanopoulos, R. Alcock, Y. Manolopoulos, Featurebased classification of timeseries data. Int. J. Comput. Res. 10(3), 49–61 (2001)
J. Ni, R.H. Drieberg, P.I. Rockett, The use of an analytic quotient operator in genetic programming. IEEE Trans. Evol. Comput. 17(1), 146–152 (2012)
M. Nicolau, A. Agapitos, Understanding grammatical evolution: Grammar design. In: Handbook of Grammatical Evolution, Springer, pp 23–53 (2018)
R.M. Nosofsky, Attention, similarity, and the identificationcategorization relationship. J. Exp. Psychol: General 115(1), 39 (1986)
M. O’Neill, C. Ryan, Grammatical evolution. In: Grammatical evolution, Springer, pp 33–47 (2003)
C. Phua, V. Lee, K. Smith, R. Gayler, A comprehensive survey of data miningbased fraud detection research. arXiv preprint arXiv:10096119 (2010)
R. Poli, WB. Langdon, NF. McPhee, JR. Koza, A field guide to genetic programming. Lulu. com (2008)
T. Rakthanmanon, B. Campana, A. Mueen, G. Batista, B. Westover, Q. Zhu, J. Zakaria, E. Keogh, Addressing big data time series: mining trillions of time series subsequences under dynamic time warping. ACM Trans. Knowl. Discov. Data (TKDD) 7(3), 10 (2013)
C.A. Ratanamahatana, E. Keogh, Making timeseries classification more accurate using learned constraints. In: Proceedings of the 2004 SIAM international conference on data mining, SIAM, pp 11–22 (2004)
T. Schlegl, P. Seeböck, SM. Waldstein, U. SchmidtErfurth, G. Langs, Unsupervised anomaly detection with generative adversarial networks to guide marker discovery. In: International Conference on Information Processing in Medical Imaging, Springer, pp 146–157 (2017)
T. Schreiber, A. Schmitz, Discrimination power of measures for nonlinearity in a time series. Phys. Rev. E 55(5), 5443 (1997)
A. Shifaz, C. Pelletier, F. Petitjean, G.I. Webb, Ts. Chief, Webb GI, Chief Ts, A scalable and accurate forest algorithm for time series classification. Data Min. Knowl. Discov. 34(3), 742–745 (2020)
H.J. Shin, D.H. Eom, S.S. Kim, Oneclass support vector machinesan application in machine fault detection and classification. Comput Ind. Eng. 48(2), 395–408 (2005)
M. Virgolin, T. Alderliesten, P.A. Bosman, On explaining machine learning models by evolving crucial and compact features. Swarm Evol. Comput. 53, 100640 (2020)
F. Wilcoxon, Individual comparisons by ranking methods. Biometrics Bullet 1(6), 80–83 (1945)
S.J. Wilson, Data representation for time series data mining: time domain approaches. Wiley Interdiscip. Rev: Comput. Stat. 9(1), e1392 (2017)
L. Ye, E. Keogh, Time series shapelets: a novel technique that allows accurate, interpretable and fast classification. Data Min. Knowl. Discov. 22(1–2), 149–182 (2011)
Acknowledgements
This work is funded by ICON plc. We would like to thank the reviewers for their thoughtful comments and efforts towards improving our manuscript.
Funding
Open Access funding provided by the IReL Consortium.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Area Editor: UnaMay O’Reilly.
Appendix A examples of featureextractors
Appendix A examples of featureextractors
Below are listed the featureextractors used to produce Fig. 9. In the text T represents a time series and L its length.

Feature 1

Class (1) \(\texttt {Min(T[1:59])+CE(T[1:59])}\)

Class (2) \(\texttt {AQ(FFT(T[1:57]),Kurtosis(T[1:57]))}\)

Class (3) \(\texttt {AQ(Mean(T[9:L])*STD(T[9:L])Max(T[9:L]),}\)
\(\texttt {STD(T[9:L]))}\)

Class (4) \(\texttt {Min(T[4:34])}\)

Class (5) \(\texttt {Mean(T[1:50])*AQ(Min(T[1:50]),Mean(T[1:50]))}\)

Class (6) \(\texttt {AQ(Below0(T[36:L]),LinearTrend(T[36:L])}\)
\(\texttt {*Min(T[36:L])*Median(T[36:L])TRAS(T[36:L])}\)


Feature 2

Class (1) \(\texttt {Above0(T[1:53])AQ(AR(T[1:53]),Max(T[1:53]))}\)

Class (2) \(\texttt {AQ(FFT(T[1:34]),AQ(AR(T[1:34])+FFT(T[1:34])),}\)
\(\texttt {AbsSum(T[1:34])*AQ(FFT(T[1:34]),AQ(AR(T[1:34])+}\)
\(\texttt {LinearTrend(T[1:34])*LinearTrend(T[1:34]),TRAS(T[1:34])}\)

Class (3) \(\texttt {AQ(Max(T[1:34]),AQ(Min(T[1:34]),AQ(Mean(T[1:34])}\)
\(\texttt {*Skewness(T[1:34])+Max(T[1:34])*AR(T[1:34]),}\)
\(\texttt {Median(T[1:34]))))}\)

Class (4) \(\texttt {Above0(T[22:L])AQ(AQ(AQ(Above0(T[22:L]),}\)
\(\texttt {Autocorrelation(T[22:L])),AbsSum(T[22:L])}\)
\(\texttt {LinearTrend(T[22:L])+Min(T[22:L])),Mean(T[22:L]))}\)
\(\texttt {AQ(Max(T[22:L]),Min(T[22:L])+LinearTrend(T[22:L])}\)

Class (5) \(\texttt {Max(T[7:L])*Mean(T[7:L])}\)

Class (6) \(\texttt {Below0(T[7:22])AQ(AR(T[7:22])*Max(T[7:22]),}\)
\(\texttt {AQ(AQ(AQ(Min(T[7:22])*Autocorrelation(T[7:22]),}\)
\(\texttt {Below0(T[7:22])),AQ(MeanChanges(T[7:22]),CE(T[7:22]))),}\)
\(\texttt {CE(T[7:22]))AbsSum(T[7:22])+Mean(T[7:22]))}\)

Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Mauceri, S., Sweeney, J., Nicolau, M. et al. Feature extraction by grammatical evolution for oneclass time series classification. Genet Program Evolvable Mach 22, 267–295 (2021). https://doi.org/10.1007/s1071002109403x
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s1071002109403x