Yes, I think your right - I just re-read and there is mention that the CNN is performed using a GPU. So as you say this is a perfect fit for Akida and Akida would probably do all these steps in one and be an easier implementation and one would assume be even more power efficient. The description of the learning is interesting though, it does describe a much briefer learning process than one would expect via a CNN, so I'm not sure how that works.
Akida Advantages, page-42
-
- There are more pages in this discussion • 2 more messages in this thread...
You’re viewing a single post only. To view the entire thread just sign in or Join Now (FREE)
Add BRN (ASX) to my watchlist
 (20min delay)
|
|||||
|
Last
19.5¢ |
Change
0.005(2.63%) |
Mkt cap ! $361.9M | |||
| Open | High | Low | Value | Volume |
| 19.5¢ | 20.0¢ | 19.0¢ | $661.1K | 3.388M |
Buyers (Bids)
| No. | Vol. | Price($) |
|---|---|---|
| 14 | 516291 | 19.5¢ |
Sellers (Offers)
| Price($) | Vol. | No. |
|---|---|---|
| 20.0¢ | 1158060 | 31 |
View Market Depth
| Last trade - 16.10pm 17/07/2024 (20 minute delay) ? |
| BRN (ASX) Chart |