Berkeley Brain Theories Workshop & More
This month, I’m pleased to share details about an upcoming event, a new article, and a partner update. First, Jeff will be speaking at the Computational Theories of the Brain Workshop hosted by the UC Berkeley Simons Institute for the Theory of Computing, April 16-18. The workshop is part of a semester-long series aimed at bridging the gap between brain science and theoretical computer science. Researchers from both areas have been invited to participate in the hopes of tackling some of the most challenging problems in brain science today.
Jeff’s talk, “Does the neocortex use grid cell-like mechanisms to learn the structure of objects?” will focus on a proposal that the neocortex learns models of objects in a similar method that the entorhinal cortex uses to map to environments. Jeff first spoke about this idea at a talk he gave at MIT in December, but he will be elaborating on it and incorporating new insights. Registration is still open, so if you’ll be in the Bay Area this April, reserve your spot today. We’ll share the video recording of Jeff’s talk after the event. In the meantime, if you want to learn more about grid cells, watch the latest HTM School video on the topic.
Also, we published a new article that I encourage you to read, “The Thousand Brains Model of Intelligence,” which expands on one of the major implications that came out of our latest paper. In that paper, “A Theory of How Columns in the Neocortex Enable Learning the Structure of the World,” we stated that a single cortical column can learn complete objects through movement, making it much more powerful than previously assumed. The Thousand Brains Model of Intelligence suggests that if every cortical column is able to learn complete objects, then our brain is not building one model of the world. It’s building thousands of models in parallel! If our proposal is right, this will affect not only neuroscience but also the future of AI. While the paper was deeply technical and scientific, the Thousand Brains piece is intended for a general audience.
Lastly, in partner news, Erik Graf, Head of R&D at Cortical.io will be speaking at the European Conference on Information Retrieval March 26-29 in Grenoble, France. His talk, “AI in the Wild” will be an overview of lessons the company has learned while helping their customers tackle business problems through AI. Similarly, they posted a new piece on how their semantic fingerprints are helping customers overcome rapidly evolving language challenges with two key components of intelligence: adaptiveness and versatility.
Thank you for continuing to follow Numenta.