Weakly Supervised Learning for Activity Recognition from Time Series Data Public Deposited

http://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/5h73q128v

Descriptions

Attribute NameValues
Creator
Abstract or Summary
  • The thesis focuses on activity recognition from sensor data, which has spurred a great deal of interest due to its impact on health care and security. Previous work on activity recognition from multivariate time series data has mainly applied supervised learning techniques which require a high degree of annotation effort to label the training data with the start and end times of each activity. Multi-instance learning (MIL) and multi-instance multi-label learning (MIML) are weakly supervised learning alternatives to the standard supervised learning framework, in which ambiguity in the labeling can reduce the annotation effort needed to produce labeled training data. We introduce generative graphical models for MIL and MIML based on auto-regressive processes. Our first work, based on a mixture of auto-regressive processes, assumes that instances within a bag are independent. We then relax the i.i.d. assumption for instances and extend the model by considering the sequential structure within a bag. Finally we take a MIML approach to predict the presence of multiple activities within a time interval. For each approach, we evaluate against other state-of-the-art algorithms to demonstrate the effectiveness of the proposed approach on multiple real-world data sets.
Resource Type
Date Available
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Non-Academic Affiliation
Keyword
Rights Statement
Publisher
Peer Reviewed
Language
Replaces
Additional Information
  • description.provenance : Approved for entry into archive by Julie Kurtz(julie.kurtz@oregonstate.edu) on 2017-06-17T23:55:31Z (GMT) No. of bitstreams: 1GuanXinze2017.pdf: 11281993 bytes, checksum: 58c2b7c16e7eb94b797da753450218a2 (MD5)
  • description.provenance : Submitted by Xinze Guan (guanx@oregonstate.edu) on 2017-06-16T16:57:28ZNo. of bitstreams: 1GuanXinze2017.pdf: 11281993 bytes, checksum: 58c2b7c16e7eb94b797da753450218a2 (MD5)
  • description.provenance : Made available in DSpace on 2017-06-21T20:12:07Z (GMT). No. of bitstreams: 1GuanXinze2017.pdf: 11281993 bytes, checksum: 58c2b7c16e7eb94b797da753450218a2 (MD5) Previous issue date: 2017-06-16
  • description.provenance : Approved for entry into archive by Steven Van Tuyl(steve.vantuyl@oregonstate.edu) on 2017-06-21T20:12:07Z (GMT) No. of bitstreams: 1GuanXinze2017.pdf: 11281993 bytes, checksum: 58c2b7c16e7eb94b797da753450218a2 (MD5)

Relationships

Parents:

This work has no parents.

Last modified

Downloadable Content

Download PDF

Items