TY - JOUR
T1 - A Scalable Neuromorphic Architecture to Efficiently Compute Spatial Image Filtering of High Image Resolution and Size
AU - Abarca, Marco
AU - Sanchez, Giovanny
AU - Garcia, Luis
AU - Avalos, Juan Gerardo
AU - Frias, Thania
AU - Toscano, Karina
AU - Perez-Meana, Hector
N1 - Publisher Copyright:
© 2003-2012 IEEE.
PY - 2019/2
Y1 - 2019/2
N2 - In this work, we propose a spiking P neuron whichis capable of performing spatial filtering operations by using new variants of the spiking neural P systems, such as synaptic weights and rules on the synapses. The inclusion of these variants have allowed us to create a compact spiking P neuron with minimal number of synapses and low computational complexity of the spiking rules. In addition, we propose a multi-FPGA neuromorphic system to support an array of very large-scale spiking P neurons to process high image resolution at high processing speeds. These neurons can be simulated by using a scalable configurable parallel hardware architecture, where its basic processing unit is a single spiking P neuron. Our results show that the proposed architecture is up to 54 and 12 times faster when compared to advanced Graphical Processing Units (GPU) and high performance CPUs, respectively. On the other hand, our proposal is 55x103 times faster than the best of existingFPGA-based neuromorphic solution.
AB - In this work, we propose a spiking P neuron whichis capable of performing spatial filtering operations by using new variants of the spiking neural P systems, such as synaptic weights and rules on the synapses. The inclusion of these variants have allowed us to create a compact spiking P neuron with minimal number of synapses and low computational complexity of the spiking rules. In addition, we propose a multi-FPGA neuromorphic system to support an array of very large-scale spiking P neurons to process high image resolution at high processing speeds. These neurons can be simulated by using a scalable configurable parallel hardware architecture, where its basic processing unit is a single spiking P neuron. Our results show that the proposed architecture is up to 54 and 12 times faster when compared to advanced Graphical Processing Units (GPU) and high performance CPUs, respectively. On the other hand, our proposal is 55x103 times faster than the best of existingFPGA-based neuromorphic solution.
KW - FPGA
KW - Spatial image filtering
KW - Spiking neural P systems
UR - http://www.scopus.com/inward/record.url?scp=85084278736&partnerID=8YFLogxK
U2 - 10.1109/TLA.2019.9082245
DO - 10.1109/TLA.2019.9082245
M3 - Artículo
SN - 1548-0992
VL - 18
SP - 327
EP - 335
JO - IEEE Latin America Transactions
JF - IEEE Latin America Transactions
IS - 2
M1 - 9082245
ER -