Abstract
This paper presents a massively parallel and scalable neuromorphic cortex simulator designed for simulating large and structurally connected spiking neural networks, such as complex models of various areas of the cortex. The main novelty of this work is the abstraction of a neuromorphic architecture into clusters represented by minicolumns and hypercolumns, analogously to the fundamental structural units observed in neurobiology. Without this approach, simulating large-scale fully connected networks needs prohibitively large memory to store look-up tables for point-to-point connections. Instead, we use a novel architecture, based on the structural connectivity in the neocortex, such that all the required parameters and connections can be stored in on-chip memory. The cortex simulator can be easily reconfigured for simulating different neural networks without any change in hardware structure by programming the memory. A hierarchical communication scheme allows one neuron to have a fan-out of up to 200 k neurons. As a proof-of-concept, an implementation on one Altera Stratix V FPGA was able to simulate 20 million to 2.6 billion leaky-integrate-and-fire (LIF) neurons in real time. We verified the system by emulating a simplified auditory cortex (with 100 million neurons). This cortex simulator achieved a low power dissipation of 1.62 μW per neuron. With the advent of commercially available FPGA boards, our system offers an accessible and scalable tool for the design, real-time simulation, and analysis of large-scale spiking neural networks.
Highlights
Our inability to simulate neural networks in software on a scale comparable to the human brain (1011 neurons, 1014 synapses) is impeding our progress toward understanding the signal processing in large networks in the brain and toward building applications based on that understanding
The higher the resolution, the more even the distribution will be; this requires more logic gates, and our result shows that a 20-bit resolution is sufficient
Along with the hardware platform, we developed a simple application programming interface (API) in Python that is similar to the PyNN programming interface (Davison et al, 2008)
Summary
Our inability to simulate neural networks in software on a scale comparable to the human brain (1011 neurons, 1014 synapses) is impeding our progress toward understanding the signal processing in large networks in the brain and toward building applications based on that understanding. In addition to smaller scale systems with detailed software or hardware neural models, it is necessary to develop a hardware architecture that is capable of simulating neural networks comparable to the human brain in terms of scale, with models with an intermediate level of biological detail, that can simulate these networks quickly, preferably in real time to allow interaction between the simulation and the environment To this end, we are focusing on a hardware friendly architecture for simulating large-scale and structurally connected spiking neural networks using simple leaky integrate-and-fire (LIF) neurons. Software simulators, such as GENESIS (Bower and Beeman, 1998) and NEURON (Hines and Carnevale, 1997), are biologically accurate and model their components with differential equations and sub-millisecond time steps This approach introduces tremendous computational costs and makes it impractical for simulating large-scale neural networks. As GPUs are still performing numeric simulations, it can take hours to simulate 1 s of activity in a tiny piece of cortex (Izhikevich and Edelman, 2008)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
