I'm interested in hardware implementation of ANNs (artificial neural networks). Are there any popular existing technology implementations in form of microchips which are purpose designed to run artificial neural networks? For example, a chip which is optimised for an application like image recognition or something similar?

  • 1
    $\begingroup$ Not sure if this is what you're really looking for, but you could research "SpiNNaker" and "TrueNorth". $\endgroup$ – Byte Commander Aug 3 '16 at 16:33
  • $\begingroup$ Nothing particular, just learning what's there for now. The names you mentioned are great start. Would be great if you could post it as an answer. $\endgroup$ – kenorb Aug 3 '16 at 16:50
  • $\begingroup$ I'm no sure whether those chips meet the criterion "which are designed to run heavy deep learning computation on it" of your question. You would probably have to make it a bit broader, as those chips are inspired by biological neural networks, but e.g. TrueNorth is not mainly intended as deep learning hardware but as general alternative to standard CPUs. $\endgroup$ – Byte Commander Aug 3 '16 at 17:05
  • $\begingroup$ I read some where that Google has made chips especially for deep learning problems which they currently use it in their products also such as google prediction api. $\endgroup$ – Eka Aug 3 '16 at 17:18

In May 2016 Google announced a custom ASIC which was is specifically built for machine learningwiki and tailored for TensorFlow. It is using tensor processing unit (TPU) which is a programmable microprocessor designed to accelerate artificial neural networks.

NeuroCores, 12x14 sq-mm chips which can be interconnected in a binary tree, see: Neurogrid, a supercomputer which can provide an option for brain simulations.

TrueNorth, a neuromorphic CMOS chip produced by IBM, which has 4096 cores in the current chip, each can simulate 256 programmable silicon "neurons", giving a total of over a million neurons.

Further readings: Neuromorphic engineering, Vision processing unit, AI accelerators

As a side note, you can always use an FPGA based piece of hardware which you can implement selected genetic algorithm (GA) directly in hardware. For example the CoDi model was implemented in the FPGA based CAM-Brain Machine (CBM)2001.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.