2020 BRN Discussion, page-9095

I did a quick comparison here:

https://hotcopper.com.au/posts/46808179/single

Just to add to the original idea I presented, WBT can do spiking neural networks but they have only demonstrated MNIST - which from what I have heard from the likes of Kris Carlson (first presentation of Akida @ neruIPS), MNIST is a bit of a meme now in that EVERYONE can do MNIST... Even their accuracy on MNIST was low... Thats not even mentioning the fact that WBTs method, "An existing framework was adapted for off-line learning", just shows how unique Brainchips technology is - in that it can learn on the edge... Incremental learning on chip..

Brainchip are already so far ahead of WBT in commercialising spiking neural networks that a statement like "has been needing to turn their concepts into realities" is very irrelevant. Brainchips technology is on the precipice of being commercialised..

Maybe a technology like WBT could add value to Brainchips tech, it's certainly not make or break. The next few months where Peter finishes his proposed Akida2&3 as suggested by their documentation, WBT will be in the rear view..

I haven't read too much about WBT but I think I have a good understanding of it when it comes to comparing with Brainchips Akida