BCIG BRAINSTORMING SESSION: "What Can We Do with an Artificial Neocortex at the NIH?”

Clinical Center (Building 10) Medical Board Room (Room 2C116)

DESCRIPTION: We will attempt to have a more focused continuation of the April 2006 BCIG “On Intelligence” Book Club Meeting dialogue, which ended with agreement to pursue the following questions: (1) How can we build an artificial neocortex? (2) What are some good biomedical applications? (3) What questions can we send to author Jeff Hawkins? (4) How should we format our meeting with Jeff when he comes to the NIH in the Fall? If the idea of building a mechanized (artificial) neocortex based on present knowledge of how the real (human) neocortex works for addressing hard biomedical problems appeals to you, please look at the book, come to this BCIG Brainstorming Session, and help us pick up the dialogue. (You are welcome even if you don’t look at the book.)

ARTIFICIAL NEOCORTEX: In his book “On Intelligence”, author Jeff Hawkins gives us his perspectives on our present understanding of the structure and function of the human neocortex. He then proposes that with this understanding we develop (in an engineering sense) models of the neocortex as general-purpose intelligent machines. This is the same motivation that drove early artificial neural network (ANN) developers, i.e. “Connectionists” to study the brain and nervous system (BANS) – for inspiration to build robust intelligent computing paradigms. Early connectionists branched away from computational neuroscientists who were modeling the BANS at the level of the neuron and up through higher BANS levels. Connectionists are not so concerned about developing more accurate BANS models as the computational neuroscientists might be. Rather connectionists use the knowledge of the brain and nervous systems for inspiration in building robust, useful, general purpose computing machines that can perform practical useful tasks for society. It is in the connectionists’ spirit that we are meeting at this BCIG Brainstorming Session to ask how we go about building such machines and how we can use them to address some of the more complex biomedical research, health care delivery and administrative problems we are now facing at the NIH. (Jim DeLeo)

NIH CONTACT: Jim DeLeo, 301-496-3848 jdeleo@nih.gov

BCIG WEB SITE: www.nih-bcig.org

NIH VISITOR INFORMATION: http://www.nih.gov/about/visitor/

3:00 pm to 4:30 pm May 4, 2006

FACILITATORS: Carl Leonard and Jim DeLeo will facilitate this brainstorming session. Jim Is a computer scientist in the NIH Clinical Center. He is the chair of BCIG and Chief of the Scientific Computing Section in the NIH Clinical Center. Carl is an electrical and computer engineer on contract to the NIH Clinical Center Scientific Computing Section where he works on data warehousing, data mining, data visualization and podcasting projects with Jim. Both Jim and Carl have experience building artificial neural network systems for biomedical applications.