30.03.2016 Views

Neuromorphic Computing From Materials to Systems Architecture

Neuromorphic-Computing-Report_FNLBLP

Neuromorphic-Computing-Report_FNLBLP

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Feature <br />

Description <br />

Clock Free <br />

Scale Free <br />

Symbol Free <br />

Grid Free <br />

Dendritic Neuron <br />

Synaptic Plasticity <br />

Synaptic Path Length <br />

Dense Connectivity <br />

Modular Cortex <br />

Broadcasting <br />

Fully asynchronous <br />

Activity can vary from local <strong>to</strong> system level scales depending upon context <br />

No single neuron or synapse represents any single item/concept <br />

Small world network geometry allows feature integration from <br />

heterogeneous and non local brain areas <br />

Nonlinear signal processing via dendrites in each neuron <br />

Most synapses exhibit plasticity at various time scales (secs <strong>to</strong> hrs) <br />

Approx. constant number of hops between different brain areas <br />

Each neuron connects <strong>to</strong> between 1000-­‐10000 other neurons <br />

Six layered modular architecture that repeats across architecture <br />

Brain areas that broadcast signals (neuromodula<strong>to</strong>ry) <strong>to</strong> all other parts <br />

Table 4. <strong>Neuromorphic</strong> system level architecture features.<br />

<strong>Neuromorphic</strong> <strong>Computing</strong>: <strong>From</strong> <strong>Materials</strong> <strong>to</strong> <strong>Systems</strong> <strong>Architecture</strong> <br />

31

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!