July 25, 2019 - July 25, 2019
Artificial Intelligence Center, Menlo Park, CA
In the midst of a global gold rush for AI, a growing number of experts realize that significant parts of the state of the art may need to be revisited. While the amount of data generated grows exponentially, our ability to make sense of it seems to have reached a tangible limit when it comes to mimicking human skills like motion, vision or understanding language. By looking at the brain’s computational mechanisms, it seems that evolution favored “semantic computation” over “algebraic computation” as used in today’s computers. In other words, the human brain processes the meaning of things in the world, whereas a computer program processes the number of things in the world.
The purpose of this talk is to present a concrete approach to systematically reverse-engineer the brain’s mechanisms for computational semantics and implement them in software. By representing natural language data in a way the human brain could actually process, many challenging computer linguistic problems like vocabulary mismatch, ambiguity or polysemy can be elegantly overcome. The Semantic Folding approach, inspired by Jeff Hawkins’ hierarchical temporal memory (HTM) theory, has since its inception in 2011 been translated into many different solutions solving key business challenges in application patterns like semantic search, semantic classification, semantic filtering and, in general, matching of unstructured data.
Speaker: Francisco Webber
Title: Semantic Supercomputing – Natural Language Understanding Beyond Statistics
Time: 4 pm
