Dynamic Time Warping
Dynamic Time Warping (DTW) is a popular technique for optimally aligning two time-dependent sequences.The technique was originally used to compare different speech patterns in automatic speech recognition. The word Dynamic in DTW refers to the fact that dynamic programming is used to solve the optimization problem of finding the minimal cost path. The approach of dynamic programming solves an optimization problem by breaking it up into smaller pieces. The global solution of the original problem emerges through the solutions of the smaller more easily solved problems.
The input data used for the DTW workspace node consists of two time-dependent sequences also known as trajectories. The DTW workspace node has numerous options providing flexibility in the way to optimally align the two time sequences. The DTW workspace node supports symmetric and asymmetric alignments along with global constraints such as the Sakoe-Chiba Band and the Itakura parallelogram, and many more additional options.