Unsupervised Activity Discovery and Characterization from Event-Streams
Abstract
We present a framework to discover and characterize different classes of everyday activities from event-streams. We begin by representing activities as bags of event n-grams. This allows us to analyze the global structural information of activities, using their local event statistics. We demonstrate how maximal cliques in an undirected edge-weighted graph of activities, can be used for activity-class discovery in an unsupervised manner. We show how modeling an activity as a variable length Markov process, can be used to discover recurrent event-motifs to characterize the discovered activity-classes. We present results over extensive data-sets, collected from multiple active environments, to show the competence and generalizability of our proposed framework.
Cite
Text
Hammid et al. "Unsupervised Activity Discovery and Characterization from Event-Streams." Conference on Uncertainty in Artificial Intelligence, 2005.Markdown
[Hammid et al. "Unsupervised Activity Discovery and Characterization from Event-Streams." Conference on Uncertainty in Artificial Intelligence, 2005.](https://mlanthology.org/uai/2005/hammid2005uai-unsupervised/)BibTeX
@inproceedings{hammid2005uai-unsupervised,
title = {{Unsupervised Activity Discovery and Characterization from Event-Streams}},
author = {Hammid, Rafay and Maddi, Siddhartha and Johnson, Amos Y. and Bobick, Aaron F. and Essa, Irfan A. and Jr., Charles Lee Isbell},
booktitle = {Conference on Uncertainty in Artificial Intelligence},
year = {2005},
pages = {251-258},
url = {https://mlanthology.org/uai/2005/hammid2005uai-unsupervised/}
}