Sometimes I miss the bleeding obvious and today while on my heart rehab walk I was doing a review of my investments and it occurred to me as I mulled over the last two years of Brainchip achievements that I had overlooked something which is in my opinion quite profound and which has been staring me in the face since last year (2021) when the successful testing of the AKD1000 commercial chip was announced.
At the time of the announcement and subsequently Brainchip has stated that the AKD1000 commercial chip tested better than expected.
Great I thought at the time but what I failed to connect until an hour ago was the next piece of information released by Brainchip.
Since Peter van der Made stated at the Brain Inspired Congress that the AKD1000 engineering sample had successfully detected Covid-19 with 94% accuracy Brainchip has recently stated that the AKD1000 chip is now detecting Covid-19 with greater than 98% accuracy. An improvement of over 4% which I contend is a consequence of the improvements made to the AKD1000 commercial chip which caused the four months delay and which has allowed Brainchip to announce that it has tested better than expected.
So what some might say but if like myself you have been a student of Brainchip and the AKIDA technology you will know that just like the human brain what AKD1000 does is recognise repeating patterns both seen and unseen. When AKD1000 receives the NaNose sensor data it is like the human brain recognising the patterns in that data which coincide with Covid-19 or any other virus which the NaNose sensor has detected.
New investors should not take my word for this however so I have extracted two questions and answers from an interview conducted by Alan Kohler with the former CEO Mr. Dinardo:
"A random thought just popped into my head. Could you mine bitcoins more cost effectively and without using so much energy? Would your chip do that better?
Well, it’s a really good question – and I always hate saying it’s a good question because I know that’s why you asked it. The difference between computing, and even in a GPU or a CPU, because we work like the brain, we’re not doing math. Your brain doesn’t really do 1 plus 1 equals 2. Your brain has a spike math that recognises I learned this when I was a kid, 1 plus 1 equals 2. We don’t do computation. That’s why it’s extremely low power. What we could do in crypto or just blockchain generally, we will never be the solution to calculate the transaction, 1 plus 1 equals 2. What we can do and what the technology is exceptional at is recognising repeating(sic) patterns.
So as this crypto environment matures or blockchain more generally, because it’s not just with respect to currency, blockchain is being applied to many other avenues as well, is we can recognise repeating patterns. We could do that for Blockchain and we could say, if there was a very high density of transactions with a very similar pattern being executed, we could do it for cyber security where we could recognise patterns for denial of service attacks, malware attacks.
More generally, harkening back to crypto, fintech or financial technology generally. High frequency trading patterns, people want to know what those patterns are so they can forecast what their future transaction should be or they want to intersect somebody else that’s trying to do exactly the same thing. So, the technology is not about compute, it’s about recognising repeating patterns.
Just before we get onto some other matters, another thought that occurs to me is, what if we could combine your technology and Peter’s ideas with quantum computing?
That could be huge. There’s always going to be kind of a side by side environment. Even if we look at today, some of our engagements because of this ability to do vision exceptionally well – we’re being approached by major automobile manufacturers around the world. Detroit, Germany… Again, we’re never going to do the compute, we’re going to recognise very quickly and very efficiently a pattern. So, we likely will sit side by side during the short to mid term, we’ll sit side by side with a CPU or a GPU, they’ll be crunching numbers and we’ll be functioning like a brain doing the thinking”
Now most here will probably have worked out the point I am making but I will state it anyway.
As AKD1000 detects repeating patterns and the move from the AKD1000 engineering samples to the AKD1000 commercial chip has improved Covid-19 pattern detection by 4% this same improvement must also have occurred in respect of every other one of AKD1000 commercial chips performance stats from object and keyword spotting to everyone of the five senses.
So think about that for a moment the already state of the art performances across these areas has now been further improved by 4% which is huge and this is before Brainchip moves to AKD2000 with LSTM.
As a side note if you have not listened too or read the full interview between Alan Kohler and Mr. Dinardo it is well worthwhile even though it is from some time ago.
My anonymous opinion only so DYOR
FF
AKIDA BALLISTA
(20min delay)
