Apple + neuromorphic, page-134

Might sound corny but my inspiration has come from watching my 9 month old grandson, who I had not seen previously due to Covid, engage in the process of learning over the last 7 days. The speed of take up of knowledge compared with traditional deep learning is extraordinary.

Then I thought to myself what if you connected the 64 AKD1000's as it is designed to do how much would it be capable of learning via one shot and incremental learning in the same time as it took Mr. Ng to run 10 million videos past his neural network. Better still what about the 1024 AKD1000's which can with certain modification be connected.

Then consider if I tell my grandson in one of his books that the animal I showed him was an Ocelot but having checked it is a CusCus it struck me that if he was taught by deep learning I would have to send him back to the factory to be completely retrained.

Instead just like AKD1000 I can re-lable the animal correctly and he will be right to go immediately but being human he might need a second shot.

Sometimes the simplest thing which you thought you understood, needs an experience or observation for its significance to be made clear.

If Mr.Ng has trained his neural network with a video made by me where I called a CusCus an Ocelot he would have to send it back for the 10 million videos to be retrained to fix the problem. Probably cheaper to make a donation to the animal naming people and switch their names they won't know or care.

My opinion only DYOR
FF

AKIDA BALLISTA