Today machine learning is largely about statistical pattern discovery and function approximation from large volumes of data. But as computing devices that interact with us in natural language become ubiquitous (e.g., Siri, Alexa, Google Home), and as computer perceptual abilities become more accurate, they open an exciting possibility of enabling end-users to teach machines similar to the way in which humans teach one another. Natural language conversations, gesturing, demonstrations, teleoperation and other modes of communication offer a new paradigm for machine learning through instruction from humans. In this talk I will discuss our effort and progress at CMU to build the next generation conversational agent that can learn from explicit verbal instruction and demonstration.
Igor Labutov's interests are in building machine learning algorithms that can learn from natural human supervision, such as verbal or visual instructions. Most recently he was a Postdoc at Carnegie Mellon Machine Learning department working with Tom Mitchell on problems of machine learning from flexible and natural forms of instruction. Prior to that he obtained his PhD from Cornell University where he was advised by Hod Lipson and Christoph Studer, and where he was a recipient of the NSF Graduate Fellowship. His undergraduate degree is from The City College of New York. In June 2018, together with Bishan Yang (also former CMU postdoc) they co-founded LAER AI., a startup based in New York City, focusing on developing next-generation semantic search tools for enterprises.