*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Samuel Shapero
Ph.D. Thesis Proposal
Technology Square Research Building, Room 523
Advisor: Paul Hasler (Georgia Institute of Technology)
Committee Members:
Dr. David Anderson (Georgia Institute of Technology)
Dr. Chris Eliasmith (University of Waterloo)
Dr. Christopher Rozell (Georgia Institute of Technology)
Dr. Garrett Stanley (University of Washington)
Sparse coding is a Bayesian optimization program with application in compressed sensing and modeling of sensory cortical systems. Previous sparse coding implementations have relied on digital algorithms whose power consumption scales poorly with problem size, rendering them unsuitable for low-power, portable applications. The Locally Competitive Algorithm (LCA) is a recent sparse coding method that can be implemented with analog hardware, allowing improvements in both power consumption and convergence time. The goal of this proposal is to develop a spiking version of the LCA using a network of artificial neurons in analog hardware. The proposed research will leverage floating gate technology to ensure computational accuracy and allow reprogrammability. Floating gate programming speed will be optimized to allow fast adaptation of synaptic weights. This adaptation will then be incorporated to allow the proposed network to learn receptive fields that optimally encode natural images.
