Info

此问题已关闭。 请重新打开它进行编辑或回答。

Dynamic time warping with conditions

2 次查看（过去 30 天）
Juan Estrella-Martínez
关闭： Juan Estrella-Martínez 2020-7-27
Hello all.
I just discovered the joys of the dtw function in Matlab. However, the intial results leave a bit to be desired. I'll explain my problem:
I am trying to match sequential data to reference data. The sequential data S with reference data R
S=[S1,S2,S3,...,Si]
% Usually i=1:10 but sometimes i=1:3, 1:4, etc.
R=[R1,R2,R3,...,Rj]
% j=1:13 is ALWAYS true
I want to carry out a dynamic time warping such that I obtain two new vectors, the matched j-index for S (WS) and an uncertainty (U) measured in "number of elements to the left and to the right of the match" that would have been good matches too.
Example:
WS=[2,3,5,5,6,6,7,8,10,12]
U=[0,1,2,1,2,2,1,0,1,2]
I need the warping to obey certain rules
  1. S1 is not necessarily matched with R1
  2. WS is either always growing or some of the elements repeat
  3. The resulting warping of S can be described by a sigmoidal curve (that is, fitting a sigmoid to the scatter data (WS,S) will have a relatively high r-squared)
How can I impose these conditions to the dtw function?
  1 个评论
Juan Estrella-Martínez
Juan Estrella-Martínez 2020-7-27
I think the question might be badly formulated or that I can get what I need with the default function. I'm closing the quesiton.
Thanks for tyhe interest!

此问题已关闭。

回答（0 个）

此问题已关闭。

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by