Dynamic Programming
Analysis of sibling time series data: alignment and difference detection
Analysis of sibling time series data: alignment and difference detection
Toward accurate dynamic time warping in linear time and space
Intelligent Data Analysis
A segment-wise time warping method for time scaling searching
Information Sciences: an International Journal
A combined alignment and registration scheme of lesions with psoriasis
Information Sciences: an International Journal
Classification and alignment of gene-expression time-series data
Classification and alignment of gene-expression time-series data
Dynamic time warping constraint learning for large margin nearest neighbor classification
Information Sciences: an International Journal
| Hi-index | 0.07 |
Signal alignment is one of the most commonly used strategies in analyzing a group of time series in order to learn the variations or common patterns across individual signals. A pairwise alignment algorithm aligns two signals by warping the time axis of the first signal so that the warped signal is ''near'' to the second. The majority of alignment algorithms are focused on extracting features like the locations of significant peaks or peak widths, and using those features in aligning the signals instead of raw signal. Although this approach allows fast alignments, it suffers from the risk of missing important features, leading to inaccurate alignments. In this paper, a novel Signal Alignment method based on Genetic Algorithm (SAGA) is proposed to align raw signals by first modeling the warping function with an ODE model. The parameters of the warping function are then optimized by using a genetic algorithm. The SAGA does not require feature extraction and it preserves the smoothness of the signals. The performance of the proposed method is evaluated on two sets of synthetic and real world datasets and compared to the well-known alignment algorithms. The results show that SAGA is a powerful algorithm that can compete with the others.