470 likes | 529 Views
Lecture 10.1 Time Series (Dis)Similarity (Dynamic Time Warping). CMSC 818W : Spring 2019. Tu-Th 2:00-3:15pm CSI 2118. Nirupam Roy. Apr. 25 th 2019. Similarity between two time series. Signal A:. Signal B:. Similarity between two time series: Euclidian distance.
E N D
Lecture 10.1 Time Series (Dis)Similarity (Dynamic Time Warping) CMSC 818W : Spring 2019 Tu-Th 2:00-3:15pm CSI 2118 Nirupam Roy Apr. 25th 2019
Similarity between two time series Signal A: Signal B:
Similarity between two time series: Euclidian distance Sqrt((a1-b1)2 + (a2-b2)2 ) (a1, a2) (b1, b2)
Similarity between two time series: Euclidian distance Signal A: N number of points Signal B: Euclidian distance:
Similarity between two time series: Euclidian distance Signal A: N number of points Signal B: Euclidian distance:
Similarity between two time series: Euclidian distance Euclidian distance
Similarity between two time series: Euclidian distance Desired approach
Similarity between two time series: Euclidian distance Desired approach
Dynamic Time Warping Slides by Quim LlimonaTorrashttps://lemonzi.files.wordpress.com/2013/01/dtw.pdf
DTW : An example [Citation] Slides take from Thales Sehn Körting https://www.youtube.com/watch?v=_K1OsqCicBY
|Ai-Bj| + min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} |9-3| + min{ 5, 5, 11} = 6 + 5 = 11 A i B j
min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} A i B j
min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} A i B j
min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} A i B j
DTW : Recap Reference: http://www.cs.ucr.edu/~eamonn/KAIS_2004_warping.pdf
DTW : Recap |Ai-Bj| + min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} |9-3| + min{ 5, 5, 11} = 6 + 5 = 11 A i B j
min{ D[i-1,j-1], D[i-1,j], D[i,j-1]} A i B j
DTW : Complexity What is the complexity of DTW algorithm?
DTW : Constraints Slides by Quim LlimonaTorrashttps://lemonzi.files.wordpress.com/2013/01/dtw.pdf
A i B j