3
$\begingroup$

What makes neuromorphic architectures more efficient (less heat and power consumption) than von Neumann architecture for complex tasks? Except inspiration from biological systems, what are some formal results on this? Any reference to books or articles is appreciated as well.

Improve this question
$\endgroup$
4
  • $\begingroup$ arxiv.org/abs/1206.3227 $\endgroup$
    – Bug Killer
    Commented Nov 17, 2013 at 5:35
  • $\begingroup$ The idea of the paper is very interesting, thanks. $\endgroup$
    – trxw
    Commented Nov 17, 2013 at 20:01
  • $\begingroup$ in a way this is more an EE/physics question. note also youre "comparing apples & oranges" $\endgroup$
    – vzn
    Commented Nov 18, 2013 at 3:04
  • 1
    $\begingroup$ ncbi.nlm.nih.gov/pmc/articles/PMC3181466 (specifically the citations in Figure 1's caption.) The comparison is a bit silly though: they're comparing single neurons, and it isn't clear how many neurons vs. how many digital computations would be required to implement any algorithm other than "simulate the Hodgkin-Huxley system of nonlinear differential equations." To do a real comparison you need to compare a digital circuit of equally low precision and implemented with something like adiabatic charge recovery logic. $\endgroup$
    – Wandering Logic
    Commented Nov 18, 2013 at 21:42

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.