INSIGHT: Efficient and Effective Instance Selection for Time-Series Classification

Abstract

Abstract. Time-series classification is a widely examined data mining task with various scientific and industrial applications. Recent research in this domain has shown that the simple nearest-neighbor classifier using Dynamic Time Warping (DTW) as distance measure performs exceptionally well, in most cases outperforming more advanced classification algorithms. Instance selection is a commonly applied approach for improving efficiency of nearest-neighbor classifier with respect to classification time. This approach reduces the size of the training set by selecting the best representative instances and use only them during classification of new instances. In this paper, we introduce a novel instance selection method that exploits the hubness phenomenon in time-series data, which states that some few instances tend to be much more frequently nearest neighbors compared to the remaining instances. Based on hubness, we propose a framework for score-based instance selection, which is combined with a principled approach of selecting instances that optimize the coverage of training data. We discuss the theoretical considerations of casting the instance selection problem as a graph-coverage problem and analyze the resulting complexity. We experimentally compare the proposed method, denoted as INSIGHT, against FastAWARD, a state-of-the-art instance selection method for time series. Our results indicate substantial improvements in terms of classification accuracy and drastic reduction (orders of magnitude) in execution times.

Fast Classification of Electrocardiograph Signals via Instance Selection

“…In this paper we exploit hubness for instance selection for time series classification algorithms. In our previous work [35] we proved that instance selection is an NP-complete problem and discussed coverage of the selected instances. Here, in contrast, we focus on hub-based instance selection for electrocardiography.…”

Classification of Electroencephalograph Data: A Hubness-aware Approach

“…limiting the warping window size), indexing and reducing the length of the time series used. For more details we refer to [32] and the references therein. On the other hand, we note that distances between different pairs of training instances can be calculated independently, therefore, computations can be parallelized and implemented on a distributed supercomputer (cloud).…”

“…We refer to this phenomenon as the presence of hubs or hubness for short, and the classifiers that take this phenomenon into account are called hubness-aware classifiers. Hubness-aware classifiers were originally proposed for vector data and image data [29], [30], [31], and only few works considered hubness-aware classification of time series [32], [33], [34], but none of them considered hubnessaware classifiers for EEG data.…”

Classification of gene expression data: A hubness-aware semi-supervised approach

“…[5], [12], [13], [15], [22], [25], [26], [20] and [21] for a sur-vey. These algorithms try to recognize bad hubs and reduce their influence on classifications of unlabeled instances.…”