DATE: | Thu, Feb 5, 2015 (unusual day) |
TIME: | 12:00 pm |
PLACE: | Council Room (SITE 5-084) |
TITLE: | Lifelong Machine Learning and Reasoning |
PRESENTER: | Danny Silver Acadia University |
ABSTRACT:
Lifelong Machine Learning (LML) considers intelligent systems that
learn many tasks over a lifetime, accurately and efficiently retaining
the knowledge they have learned and using that knowledge to more
quickly and accurately learn new tasks. In this tutorial we will
review a framework for LML, define its requirements, and present
solutions for the key problems that involve knowledge consolidation
and transfer learning using multiple task learning methods.
Opportunities for advances in artificial intelligence lie at the locus
of machine learning and knowledge representation; specifically,
knowledge consolidation can provide insights into common knowledge
representation for use in learning and reasoning. With this in mind,
the final part of the talk will discuss recent work on extending LML
to the learning of common background knowledge for the purposes of reasoning
(this extension we call Lifelong Machine Learning and Reasoning, or LMLR).
|