Active visual category learning

Show simple record

dc.contributor.advisor Grauman, Kristen Lorraine, 1979-
dc.creator Vijayanarasimhan, Sudheendra 2011-06-02T14:33:43Z 2011-06-02T14:34:23Z 2011-06-02T14:33:43Z 2011-06-02T14:34:23Z 2011-05 2011-06-02 May 2011
dc.description.abstract Visual recognition research develops algorithms and representations to autonomously recognize visual entities such as objects, actions, and attributes. The traditional protocol involves manually collecting training image examples, annotating them in specific ways, and then learning models to explain the annotated examples. However, this is a rather limited way to transfer human knowledge to visual recognition systems, particularly considering the immense number of visual concepts that are to be learned. I propose new forms of active learning that facilitate large-scale transfer of human knowledge to visual recognition systems in a cost-effective way. The approach is cost-effective in the sense that the division of labor between the machine learner and the human annotators respects any cues regarding which annotations would be easy (or hard) for either party to provide. The approach is large-scale in that it can deal with a large number of annotation types, multiple human annotators, and huge pools of unlabeled data. In particular, I consider three important aspects of the problem: (1) cost-sensitive multi-level active learning, where the expected informativeness of any candidate image annotation is weighed against the predicted cost of obtaining it in order to choose the best annotation at every iteration. (2) budgeted batch active learning, a novel active learning setting that perfectly suits automatic learning from crowd-sourcing services where there are multiple annotators and each annotation task may vary in difficulty. (3) sub-linear time active learning, where one needs to retrieve those points that are most informative to a classifier in time that is sub-linear in the number of unlabeled examples, i.e., without having to exhaustively scan the entire collection. Using the proposed solutions for each aspect, I then demonstrate a complete end-to-end active learning system for scalable, autonomous, online learning of object detectors. The approach provides state-of-the-art recognition and detection results, while using minimal total manual effort. Overall, my work enables recognition systems that continuously improve their knowledge of the world by learning to ask the right questions of human supervisors.
dc.format.mimetype application/pdf
dc.language.iso eng
dc.subject Active learning
dc.subject Object recognition
dc.subject Object detection
dc.subject Cost-sensitive learning
dc.subject Multi-level learning
dc.subject Budgeted learning
dc.subject Large-scale active learning
dc.subject Live learning
dc.subject Machine learning
dc.subject Visual recognition system
dc.subject Artificial intelligence
dc.title Active visual category learning 2011-06-02T14:34:23Z
dc.contributor.committeeMember Dhillon, Inderjit S.
dc.contributor.committeeMember Aggarwal, J K.
dc.contributor.committeeMember Mooney, Raymond J.
dc.contributor.committeeMember Torralba, Antonio
dc.description.department Computer Sciences
dc.type.genre thesis
dc.type.material text Computer Sciences Computer Science University of Texas at Austin Doctoral Doctor of Philosophy

Files in this work

Size: 11.83Mb
Format: application/pdf

This work appears in the following Collection(s)

Show simple record

Advanced Search


My Account