Configurable Hardware Integrate and Fire Neurons for Sparse Approximation (bibtex)

by S. Shapero, C.J. Rozell and P. Hasler

Abstract:

Sparse approximation is an important optimization problem in signal and image processing applications. A Hopfield-Network-like system of integrate and fire (IF) neurons is proposed as a solution, using the Locally Competitive Algorithm (LCA) to solve an overcomplete L1 sparse approximation problem. A scalable system architecture is described, including IF neurons with a non- linear firing function, and current-based synapses to provide linear computation. A network of 18 neurons with 12 inputs is implemented on the RASP 2.9v chip, a Field Programmable Analog Array (FPAA) with directly programmable floating gate elements. Said system uses over 1400 floating gates, the largest system programmed on a FPAA to date. The circuit successfully reproduced the outputs of a digital optimization program, converging to within 4.8% RMS, and an objective cost only 1.7% higher on average. The active circuit consumed 559 microamps of current at 2.4V, and converges on solutions in 25$ microseconds, with measurement of the converged spike rate taking an additional 1ms. Extrapolating the scaling trends to a N=1000 node system, the Analog LCA compares favorably with State-of-the-Art digital solutions, and analog solutions using a non-spiking approach.

Reference:

Configurable Hardware Integrate and Fire Neurons for Sparse ApproximationS. Shapero, C.J. Rozell and P. Hasler. Neural Networks, vol. 45, pp. 134–143, September 2013. Special issue on Neuromorphic Engineering: from Neural Systems to Brain-Like Engineered Systems.

Bibtex Entry:

@Article{shapero.12, author = {Shapero, S. and Rozell, C.J. and Hasler, P.}, title = {Configurable Hardware Integrate and Fire Neurons for Sparse Approximation}, abstract = {Sparse approximation is an important optimization problem in signal and image processing applications. A Hopfield-Network-like system of integrate and fire (IF) neurons is proposed as a solution, using the Locally Competitive Algorithm (LCA) to solve an overcomplete L1 sparse approximation problem. A scalable system architecture is described, including IF neurons with a non- linear firing function, and current-based synapses to provide linear computation. A network of 18 neurons with 12 inputs is implemented on the RASP 2.9v chip, a Field Programmable Analog Array (FPAA) with directly programmable floating gate elements. Said system uses over 1400 floating gates, the largest system programmed on a FPAA to date. The circuit successfully reproduced the outputs of a digital optimization program, converging to within 4.8% RMS, and an objective cost only 1.7% higher on average. The active circuit consumed 559 microamps of current at 2.4V, and converges on solutions in 25$ microseconds, with measurement of the converged spike rate taking an additional 1ms. Extrapolating the scaling trends to a N=1000 node system, the Analog LCA compares favorably with State-of-the-Art digital solutions, and analog solutions using a non-spiking approach.}, year = 2013, volume = 45, month = sep, pages = {134--143}, journal = {Neural Networks}, note = {Special issue on Neuromorphic Engineering: from Neural Systems to Brain-Like Engineered Systems.}, url = {http://www.sciencedirect.com/science/article/pii/S089360801300097X} }

Powered by bibtexbrowser