Bravo
If ARM was an arm, BRN would be its biceps💪!
@Slymeat,
For the Weebit fanciers:-
Some found facts re hybrid analog/digital NNs:
https://www.mdpi.com/2674-0729/1/2/8/pdf?version=1660304162
This paper is about self-healing Analog Front End (AFEX) and Spiking Analog Front End (SAFEX) in hybrid (analog/digital) NNs.
Akida gets a gurnsey in Fig 3 (p. 86) as an example of Spike-Based Processing, juxtaposed with Conventional Digital Processing (a microprocessor).
It also mentions Akida in passing in footnote 9 (page 84), but just bundles it in with other chips like Loihi, True North and Cerebra's wafer-sized SoC.
Of particular interest is the discussion on self-healing circuits on page 84:
"However, it must be remembered that technical systems do not have a metabolism and observed regrowth of lost cells, tissues, limbs, or even a whole entity in, e.g., axolotl salamanders or planarians, and can, as of yet, only be rivaled to the extent of the redundancy supplied at manufacturing time."
This is something which could be done fairly easily with Akida. The beauty of Akida's use of identical NPU's and nodes and the inter-node communication matrix is that, if sufficient redundant nodes were provided, the configuration MCU (Cortex) could be programmed to test the nodes and provide a detour to good nodes in the event of a node failure. It would "simply" require a reprogramming of the dud node's address to an available redundant node's address and consequential assigning of weights.
For AFEX, the paper discusses the use of a current feedback in amplifier (CFIA) circuit (Fig 8). The 3rd last par on page 91 discusses some f the problems of inaccuracy with analog CIFAR circuits.
From the closing paragraph (p.112), it does not appear that this project will bear fruit any time soon.
PS: As a child, I was fascinated by axolotls.
Knock knock
Who's there
Axolotl
Axolotl who?
you sure do Axolotl questions!