Data-selection in learning algorithms

Abstract

In recent years, the amount of information stored in data acquisition devices has increased exponentially, due to the massive multiple-input multiple-output (MIMO) antennas, social networks, and distributed networks. In this era of Big Data, we face the challenge of eficiently utilizing a large amount of data to extract valuable information. Therefore, it is essential to define criterials to decide if the data is relevant or not during the learning process. This work proposes a data selection strategy for two areas: adaptive filtering and neural networks. In both situations, data could be discarded, reducing the computational cost and, in some cases, the accuracy of the estimate. The applications analyzed in this work include synthetic and real data, these verify the efectiveness of the proposed algorithms that may achieve significant reductions in computational costs without sacrificing estimation accuracy due to the selection of the data.