Behavior Analysis through Routine Cluster Discovery in Ubiquitous Sensor Data
Citations
5 citations
References
2,233 citations
1,054 citations
"Behavior Analysis through Routine C..." refers methods in this paper
...Here A represents P (w|d), X represents P (w|t), and D represents P (t|d) which is similar to Matrix factorization as described in previous models, so LDA is also based on the same concept as the PLSA and NMF....
[...]
...From the above results, it can be seen that the NMF model is the best fit for IL data....
[...]
...As shown in Figure 1(a), for IL dataset optimal number of topics in the case of NMF Model is 80 with coherence score 0.7315....
[...]
...Perplexity and coherence are both scored for probabilistic models like LDA, PLSA, and NMF where the scores essentially represent what chances are there of discovering a new cluster....
[...]
...Using those, the best result is given by LDA which yields 20 routines in Kyoto, and NMF which yields 26 in IL, and 80 in MERL....
[...]
473 citations
"Behavior Analysis through Routine C..." refers methods in this paper
...[7] proposed a method for modeling and discovering daily routines from on-body sensor data....
[...]
468 citations
"Behavior Analysis through Routine C..." refers background in this paper
...[9] combine sequence mining and clustering algorithm to identify activities in a home environment which...
[...]
373 citations
"Behavior Analysis through Routine C..." refers methods in this paper
...For a Smart home environment, we focused on the Kyoto dataset generated by the Centre for Advanced Studies in Adaptive System(CASAS)[11] at the Washington State Universitys School of Electrical Engineering and Computer Science Department....
[...]