Open creatist opened 5 years ago
You should check the length of your time-series data. In the original paper and my experiments, when the length of each time series is less than 300 points, the speed of the FastDTW and the original DTW with JIT acceleration is almost the same. Especially when the data points are less than 50, the accelerated DTW is much faster than the FastDTW.
I tried fastdtw, very slow . Anyone has this problem?