Faculty and Student Teams Program

Project Descriptions

Los Alamos National Laboratory
High Performance Neurocomputation

Project Description

After decades of exponential growth in power, existing computer architectures still fail to match the ability of the mammalian brain to interpret, respond to, and learn from natural sensory inputs. Rapid progress in neuroscience suggests an alternative strategy for achieving brain-like behavior: identifying the computational primitives that underlie the processing in biological neural circuits. This project develops high-performance neural simulation tools and to use them to find these primitives, which make the brain so much more powerful than the familiar von Neumann computer or artificial neural networks.

This project includes two major code development efforts.  PetaVision is high-performance software for modeling cortical dynamics at a level of detail sufficient to investigate the computational primitives underlying neural computation.  PetaVision includes generic layers and neuron types in a manner that was modular, extensible, multi-threaded and platform independent. The code can model lateral cocircular synaptic interactions between neurons in the primary visual cortex.

A second effort within this program develops software for simulating the hierarchical structure of the visual cortex based on feed forward models.  This modelincludes backend classifiers for assessing performance on viewpoint invariant, object detection tasks. This model includes representations of V1 and V2/V4, simple and complex cells (and their generalizations at higher levels), imprinting, plus a range of back end classifiers of various levels of sophistication, has been implemented in C++.

Other areas of research within this project include work on appropriate patterns of lateral synaptic connectivity for extracting texture and luminance flows for defining locally smooth surfaces; models of dendritic processing based on spiking, non-linear subunits; psychophysical measures of human performance on visual cognition tasks and the role of fixational eye movements on the encoding of visual information in the form of spatiotemporal correlations between retinal neurons.

Applicants Responsibilities and Relationships to Project
Applicants will receive support under the Department of Energy Faculty Student Team Research Program (FaST) to work collaboratively with the project research team at LANL for 10 weeks during the Summer 2009. Faculty will be expected to identify students from their campus to participate in the FaST program offered by the Department of Energy at LANL. Faculty will provide mentorship and advising support to students during the summer research activities. It is desired that the Faculty member will define specific responsibilities and roles as it relates to the project and become an integral part of the research team working on this project, with the goal of supporting the project through the academic year on her or his campus.

Qualification of Ideal Candidate
Faculty: Ph.D. in computer science with experience in computational neuroscience. Works well both in a collaborative environment with researchers and also independently. Experience teaching and mentoring students. Currently teaches and collaborates with students in his/her field. Possesses good written and verbal communication skills. Willing to work at LANL for an extended period during the summer.

Student: Working towards a BS/BA in engineering or relevant science. Works well in collaboration with faculty, other students, and researchers. Possesses good written and verbal communication skills. Willing to work at LANL for an extended period.

For More Information Contact:
Scott Robbins
Education and Post-doc Office
P.O. Box 1663, MS M709
Los Alamos National Lab
Los Alamos NM 87544
srobbins@lanl.gov
505 667 3639 (O)
505 665 6871 (F)

Support and Financial Commitments

See Financial Information.