applied math seminar
Event Description:
Title: Novel approaches to numerical computing using neural algorithms
Abstract The connection between computation and the brain has been discussed for decades, but specific cases of the fields influencing one another theoretically have been limited. The need to advance this interaction has increased in recent years, especially with the arrival of brain-inspired “neuromorphic” hardware which promises low-power and rapid simulation of neural models and neural algorithms. In this talk, I will describe three such intersections. The first part of the talk will focus on how we can formalize a proposed neural code that accounts for the heterogeneity of neuron connectivity and responses. This code, which describes neural activity as a combinatorial model, is useful for describing the biological process of adult neurogenesis. The second part of the talk will show how such explicitly constructed neural circuits can then be used to accelerate key numerical kernels efficiently if implemented on neuromorphic hardware. We have developed several examples of this, including correlation and matrix multiplication. Finally, the third part of the talk will describe our approach for neural algorithms for implementing Markov random walk solutions for solving the diffusion equation.
======================
James Bradley Aimone, Ph.D.
jbaimon@sandia.gov
EDUCATION
University of California, San Diego – La Jolla, California
Doctorate of Philosophy in Neurosciences, 2009
Specialization in Computational Neuroscience
Thesis title: “Computational modeling of adult neurogenesis in the dentate gyrus”
Thesis Committee Chair: Fred H Gage, PhD
Thesis Committee Co-chair: Jeffrey L Elman, PhD
Rice University – Houston, Texas
Masters of Chemical Engineering, 2002
Bachelor of Science in Chemical Engineering, 2001
PROFESSIONAL EXPERIENCE
Sandia National Laboratories – Albuquerque, New Mexico
Data Driven and Neural Computing Department
Principal Member of Technical Staff, R&D Computer Science, January 2015 – Present
Senior Member of Technical Staff, R&D Cognitive Systems, October 2011 – January 2015
• Led multiple projects in design and analysis of large scale supercomputer simulations of biological neural circuits
• Leader of computational neuroscience research team studying neural inspired machine learning and scientific computing algorithms and architectures
• Neural Algorithms Core Lead and Deputy Principal Investigator for Hardware Acceleration of Adaptive Neural Algorithms (HAANA) internal R&D Grand Challenge. Coordinate research of ~20 staff focused on neural algorithms, computer vision, and machine learning applications in cyber security
The Salk Institute for Biological Studies – La Jolla, California
Laboratory of Genetics with Dr. Fred H. Gage
Postdoctoral Fellow, June 2009-September 2011
Computational Neuroscience PhD Student, September 2005 - June 2009
Research Assistant for Quantitative Biological Analysis, February 2002 - September 2005
PUBLICATION AND RESEARCH HIGHLIGHTS
• 50+ peer reviewed research papers published in fields ranging from electrical engineering, machine learning, molecular biology, and neuroscience
• Principal Investigator or Deputy Principal Investigator for over $16M of competitive internal Laboratory Directed Research & Development funding
• National Academy of Engineering Frontiers of Engineering Invitee
• Over 5700 research citations (per Google Scholar, August 2018)
• 10+ invited research seminars and invited conference presentations
• One patent awarded on adaptive neural algorithms, ten patent applications under consideration in neural algorithms and neural computing hardware
Event Contact
Contact Name: Pavel Lushnikov