Time Series Classification by Shapelet Dictionary Learning with SVM-Based Ensemble Classifier

Algorithm 1

Shapelet dictionary learning.

	Input: time series data set T, parameters p, K, λ, c, stopping threshold ε;
	Output: complete Shapelet dictionary S;
(1)	For l = 1, 2, …, L do
(2)	Initialize sub-dictionaries Dl, sub-coefficient Al, sub-location ql
(3)	Repeat
(4)	For k = 1, 2, …, K do
(5)	update qlk with (4)
(6)	update αlk with (5)
(7)	End For
(8)	For k = 1, 2, …, K do
(9)	update dlk with (7)
(10)	End For
(11)	Until convergence
(12)	End For
(13)	S = [ D1, D2, … DL]
(14)	Return S