Algorithm 4: Recognize Activity

Input:  Ar{a1,a2…an}, a set of activity

at, test sample

w1,w2, weight of ratio

n, select the number of training instances with the highest similarity

Output:label, label of at

1. Time←getTime(Ar)//get bt and et

2. AllRatio←Ø

3. k←SelectK(Ar)//SelectK is used to select the optimal k

4. clf←KMeans(n_clusters=k)//clustering

5. clf.fit(Time)//fitting the data

6. m←clf.predict({at.bt,at.et})//find the cluster where the test sample is located

7. for each a in Ar:

8.    if cluster(a) == m:

9.       ratio←getRatio(a,at,w1,w2)//calculate ratio

10.       AllRatio←◡ratio

11.    end if

12. end for

13. sort(AllRatio)//sort max to min

14. for i in range(n):

15.    val← getALLRatio(i)//get the first n instances of maximum ratio

16.    topk_y←◡val

17. end for

18. label←vote(topk_y)//vote for label

19. return label