Asymmetric Learning Vector Quantization for Efficient Nearest Neighbor Classification in Dynamic Time Warping Spaces
This addresses efficiency issues in time series classification for applications like signal processing or finance, though it is an incremental extension of existing methods to a new domain.
The paper tackles the high storage and computation costs of nearest neighbor classification with dynamic time warping (DTW) in time series by extending learning vector quantization (LVQ) to DTW spaces, resulting in superior performance over state-of-the-art prototype generation methods.
The nearest neighbor method together with the dynamic time warping (DTW) distance is one of the most popular approaches in time series classification. This method suffers from high storage and computation requirements for large training sets. As a solution to both drawbacks, this article extends learning vector quantization (LVQ) from Euclidean spaces to DTW spaces. The proposed LVQ scheme uses asymmetric weighted averaging as update rule. Empirical results exhibited superior performance of asymmetric generalized LVQ (GLVQ) over other state-of-the-art prototype generation methods for nearest neighbor classification.