Hi Fab86, it's just a few clicks away, but I'll oblige your...

Hi Fab86, it's just a few clicks away, but I'll oblige your request.

30 October 2015 - For ASX Release

BrainChip Achieves Milestone 2 – A Hardware-only Spiking Neural Network (SNN)

BrainChip Holdings Limited (“BrainChip” or “the Company”) is pleased to advise that it has
successfully achieved ahead of schedule its second milestone, the creation of a hardware only
version of its patented and proprietary autonomous learning Spiking Neural Adaptive
Processor (“SNAP”) technology.

This significant milestone was achieved via the development of a BrainChip SNN (Spiking
Neural Network) working on a Xilinx T2000 FPGA board (Field Programmable Gate Array).
FPGA’s are the precursor used to assess new ASIC (Application Specific Integrated Circuit)
designs before committing to a silicon-based microchip, that BrainChip anticipates its future
technology partners will develop for use in a vast array of potential products.

Peter van der Made August 2012:

Instead of programming a processor to behave like a digital neuron I came up with the solution to build a digital neuron and synapses from logic gates, and to put 15,000 of those on a single chip. I tested that idea in FPGA (programmable logic gates), and found that the simulated dynamic neural matrix learned very quickly.

http://www.kurzweilai.net/low-power-chips-to-model-a-billion-neurons

5 years is a very long time in the digital world, so I am not claiming that a ‘simulated dynamic neural matrix’ (now called ‘SNAP’) working on an FPGA in 2011 (or whenever exactly it was), is the same as BrainChip working on an FPGA in 2015. My point is that, by using logic gates on an FPGA as opposed to software, PVMD built a ‘simulated neuron and synapses’ and demonstrated that this design could not only perform AI but could do so much faster than any software competitor, thus floating the idea of a non von Neumann paradigm.

He then ‘exposed the FPGA to 10 frequecies, and the neural matirix learned to recognise those frequencies in speech patterns.’

That is a remarkable achievement and imo as impressive as the recent M2 in terms of what it essentially claims – the creation of a hardware only version of an autonomous learning chip. Keep in mind that Peter first filed for patent his autonomous learning dynamic artificial neural design in 2008.

So my related point, why has this very important event – i.e., his patent working in hardware - not been mentioned once in any of the announcements from BRN.

Happy to be corrected on any of the above.

All imo.