Events Calendar

PhD Preliminary Oral Exam – Mandana Hamidi Haines

Learning from Examples and Interactions

Human learns by interacting with other people/experts. Why not machines? As Artificial Intelligence (AI) is increasingly being applied to enhance and improve our lives, the challenge of supporting interaction with humans is becoming more apparent. In this proposal, we studied three different machine learning problems, in which learning from human interaction improved the performance of the system. The first one, is learning hierarchical policies by imitation. Given a set of expert demonstrations, our approach learns a hierarchical policy by actively selecting demonstrations and using queries to explicate their intentional structure at selected points. The second one, is the problem of active multi-label learning where thequeries are restricted to come from randomly varying subsets.This setting captures crowd sourcing scenarios where there are multiple experts with different types of expertise, and not all experts are available at all times. We generalize the framework of adaptive submodularity and prove the first near optimal approximation bound for a greedy policy for this setting. The third one is the problem of explaining the decisions of deep neural networks using human-recognizable visual and/or linguistic concepts. Our approach, called interactive naming, is based on enabling human annotators to interactively group the excitation patterns of the neurons in the critical layer of the network into groups called “visual concepts". We performed a systematic study visual concepts produced by five human annotators. We find that a large fraction of the activation maps have recognizable visual concepts, and that there is significant agreement between the different annotators about their denotations. Finally, we would like to apply active learning methods for speeding up the namings and reducing the annotator effort.

Major Advisor: Prasad Tadepalli
Minor Advisor: Sarah Emerson
Committee: Alan Fern
Committee: Weng-Keen Wong
GCR: John Dilles

Thursday, December 6, 2018 at 1:00pm to 3:00pm


Kelley Engineering Center, 1005
110 SW Park Terrace, Corvallis, OR 97331

Event Type

Lecture or Presentation

Event Topic

Research

Organization
College of Engineering, Electrical Engineering and Computer Science
Contact Name

Calvin Hughes

Contact Email

Calvin.Hughes@oregonstate.edu

Subscribe
Google Calendar iCal Outlook

Recent Activity