I wouldn’t say this is one-shot learning FF, I reckon it’s more...

First 24 Last

Share
Profile

Follow
jtardif999

1,088 Posts.

573

23/10/21

14:41

Post #: 57019846
Share

Originally posted by Fact Finder: ↑

The following link is for actual investors. I read this paper last year (cannot recall if I posted) but in purporting to herald a possible breakthrough in one shot learning read in its entirety it makes clear that this mathematical breakthrough is far from being the complete answer and a lot of further research and work on the theory has to be done and done successfully.

Even assuming it is successful it then needs to find its way into the market so at least three years???

This article makes clear that Brainchip’s commercial product with “ONE SHOT LEARNING” has the “MARGIN OF VICTORY” written all over it and my mate ‘Blind Freddie’ said to me don’t worry even I can see where this is going.

https://www.technologyreview.com/2020/10/16/1010566/ai-machine-learning-with-tiny-data/amp/

My opinion only (Blind Freddie is fictional) DYOR
FF

AKIDA BALLISTA

Expand

I wouldn’t say this is one-shot learning FF, I reckon it’s more like distilled labelling or soft labelling as it’s described - still a training process. An AI in the field isn’t going to learn something new from this approach - only based on its training is it going to be able to recognise anything at all. They may be able to improve the efficiency of training AI with this approach but imo it’s still as far away from one-shot as it was before.

BRN Price at posting: 46.0¢ Sentiment: Hold Disclosure: Held
4 Upvote

0 Great analysis

Reply