Approximate k-NN delta test minimization method using genetic algorithms: Application to time series

Abstract

In many real world problems, the existence of irrelevant input variables (features) hinders the predictive quality of the models used to estimate the output variables. In particular, time series prediction often involves building large regressors of artificial variables that can contain irrelevant or misleading information. Many techniques have arisen to confront the problem of accurate variable selection, including both local and global search strategies. This paper presents a method based on genetic algorithms that intends to find a global optimum set of input variables that minimize the Delta Test criterion. The execution speed has been enhanced by substituting the exact nearest neighbor computation by its approximate version. The problems of scaling and projection of variables have been addressed. The developed method works in conjunction with MATLAB's Genetic Algorithm and Direct Search Toolbox. The goodness of the proposed methodology has been evaluated on several popular time series examples, and also generalized to other non-time-series datasets.

Evolutionary optimization of neural networks with heterogeneous computation: study and implementation

“…Phase 1: Selection of the best inputs by means of an evolutionary computation based on the Delta Test [13].…”

“…In order to undertake the feature selection, it was very important to do without the intervention of a neural network, because in this way we became independent the selection of variables against the network topology. Until carrying out this study, we had used the optimization of the Delta Test using genetic algorithms, with regression problems with only one output [13], but this approach can be extended to classification problems with multiple outputs. The particular focus of this study did not concern this phase of the methodology.…”

Computational Efficiency: Can Something as Small as a Raspberry Pi Complete the Computations Required to Follow the Path?

“…Another algorithm which could be an alternative solution to this problem is to implement the K-nearest-neighbour search algorithm, which is one of the most known unsupervised learning algorithms used for data clustering and classification. This is explained briefly in [17] as "Nearest neighbour search is an optimization technique for finding closest points in metric spaces. Specifically, given a set of n reference points R and query point q, both in the same metric space V" [17].…”

The delta test: The 1-NN estimator as a feature selection criterion

“…Experimental evidence comparing to alternative methods is also provided. Choosing an efficient search scheme for high-dimensional tasks is mostly left as a practical matter of implementation, and several papers specifically about optimising the Delta test have also been published in the literature [13], [14], [15], [16], [17], [18], [19].…”