Digital Neuron Cells for Highly Parallel Cognitive Systems

Master Thesis (2017)
Author(s)

H. Lin (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

Rene Van Leuken – Mentor

Carlo Galuzzi – Graduation committee member

Arjan Van Genderen – Graduation committee member

A. Zjajjo – Graduation committee member (TU Delft - Signal Processing Systems)

Faculty
Electrical Engineering, Mathematics and Computer Science
Copyright
© 2017 Haipeng Lin
More Info
expand_more
Publication Year
2017
Language
English
Copyright
© 2017 Haipeng Lin
Graduation Date
31-08-2017
Awarding Institution
Delft University of Technology
Programme
Electrical Engineering | Circuits and Systems
Faculty
Electrical Engineering, Mathematics and Computer Science
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

The high level of realism of spiking neuron networks and their complexity require a considerable computational resources limiting the size of the realized networks. Consequently, the main challenge in building complex and biologically accurate spiking neuron network is largely set by the high computational and data transfer demands. In this thesis, I implement several efficient models of the spiking neuron with characteristics such as axon conduction delays and spike timing-dependent plasticity in a real-time data-flow learning network. With the performance analysis, the trade-offs between the biophysical accuracy and computation complexity are defined for the different models. The experimental results indicate that the proposed real-time data-flow learning network architecture allows the capacity of over 1,188 (max.6,300, depending on the model complexity) biophysically accurate neurons in a single FPGA device.

Files

MSc_LinHaipeng.pdf
(pdf | 2.21 Mb)
License info not available