Natural language communication has long been considered a defining characteristic of human intelligence. In humans, this communication is grounded in experience and real world context—“what” we say or do depends on the current context around us and “why” we say or do something draws on commonsense knowledge gained through experience. This talk will explore the question of how to imbue learning agents with these abilities—i.e. how to understand and generate contextually relevant natural language in service of achieving a goal. In particular, I will focus on two components of language learning shown to be vital to such tasks in humans—interactivity and environment grounding—and show how these aspects allow for more effective language-based learning agents.
Prithviraj, or Raj, is a final year PhD Candidate in the School of Interactive Computing at Georgia Institute of Technology. He spends most of his time trying to mix language with stories, games, and reinforcement learning.