Fact Finder
Top 20
Here we go, dragging me back to existential truth yet again.Let’s be honest FF. This is what you think, as you wrote it first
FF
AKIDA BALLISTA
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
FF
AKIDA BALLISTA
Fact Finder
Top 20
Only rule on a roller coaster is stay seated. It’s those who stand up who loose their head."it is entirely possible that Akida helped clinch the commercial pilot deal between nviso and Siemens"
That, was exactly my thinking as well, Diogenese!
@Labsy @BaconLover, where did you get the 18 ingredients for KFC from!
Let's try and keep things factual here, 11 herbs and spices (of which only 1 or 2, would be a mystery), the chicken and the oil, that's 13!
Unless you're including chips and sides?
On a sour note, US futures looking miserable again already..
The roller coaster continues..
FF
AKIDA BALLISTA
Rise from the ashes
Regular
Was just changing TV channels from watching footy & on channel 44 they were conducting a road test of the Merc EQXX & singing it's praises. Speeds of up to 140KPH were reached on the autobahn & the remainder of the trip was at the speed limits. They spoke of the aerodynamics & high tech materials but did not directly mention Akida but were impressed with the 1000 KM range. The show was called REV on channel 44. Most here will probably have seen this or know about it already as there has been many articles posted but I was excited to actually see it on free to air TV.
Xhosa12345
Regular
Now linked to siemans potentially
Maybe why sp had a good day, ill take it...
Maybe comments of the recent podcast have been posted, but I have to say that listening to Doug Fairbairn was magical.
Megachips have certainly picked a top man to head their operations in the US, he has clearly stated the early product pipeline, that
being, Wearables, Security and IIOT Industrial Internet of Things, so expect our IP embedded in this fields first.
The billion company that nobody had/has ever heard of, except within the Japanese market space, until now !
I had a quiet laugh to myself earlier today while chatting to my younger sister back in New Zealand, back in 2020 while I was over
at my property in NZ, we had a conversation about buying into Brainchip at 5c, today she was considering buying a small parcel, and
I said the price was currently 0.852, it was very cheap, having myself bought more at 0.95, $1.07 and $1.16 recently, I'm not concerned
with those prices whatsoever, as I was when buying at 0.039...but my advice to her was, if you weren't prepared to listen to me then, at
the 5c mark, why listen to me now at 0.852.
I just listen to our founder and all our brilliant staff, if they say it's all on track, well, that's good enough for me.
Always trust your own gut feel, not everyone shares the same risk tolerance as I and many others do, and that's ok, some investors are
more than happy working on the yield as their return, and I except that.
Cheers and goodnight from Perth.....Tech x
Megachips have certainly picked a top man to head their operations in the US, he has clearly stated the early product pipeline, that
being, Wearables, Security and IIOT Industrial Internet of Things, so expect our IP embedded in this fields first.
The billion company that nobody had/has ever heard of, except within the Japanese market space, until now !
I had a quiet laugh to myself earlier today while chatting to my younger sister back in New Zealand, back in 2020 while I was over
at my property in NZ, we had a conversation about buying into Brainchip at 5c, today she was considering buying a small parcel, and
I said the price was currently 0.852, it was very cheap, having myself bought more at 0.95, $1.07 and $1.16 recently, I'm not concerned
with those prices whatsoever, as I was when buying at 0.039...but my advice to her was, if you weren't prepared to listen to me then, at
the 5c mark, why listen to me now at 0.852.
I just listen to our founder and all our brilliant staff, if they say it's all on track, well, that's good enough for me.
Always trust your own gut feel, not everyone shares the same risk tolerance as I and many others do, and that's ok, some investors are
more than happy working on the yield as their return, and I except that.
Cheers and goodnight from Perth.....Tech x
Last edited:
Learning
Learning to the Top 🕵♂️
Very true FF,Only rule on a roller coaster is stay seated. It’s those who stand up who loose their head.
FF
AKIDA BALLISTA
A roller coaster ride is fun and exciting and sometime dangerous. But, one must stay seated and buckle up. And it will arrive at the destination!
The moment you unbuckle and get off halfway, thats is dangerous (financially).
So stay seated and enjoy the ride everyone
Its great to be a shareholder.
DingoBorat
Slim
I highly recommend anyone, who's interested in the workings of market manipulation, watch this video.
It is mainly aimed at informing people who like to trade, but provides valuable insight, into the actions of "Market Movers", "Smart Money" and "Professional Shorters".
Particularly interesting, was his statements that say, these individuals/algorithms, can see all orders in the system.
Including "stop loss" which are basically sell orders.
And the "dropping of the penny" for me, that they are responsible, for the big moves, in both directions, as traps..
These measures, apply to all markets and will be played out, on a larger scale, in the US tonight.
Incidently, I've only watched 2 or 3, of this guy's videos, a while back, but he's responsible for my adoption of Heikin-Ashi candles, when I do check the charts..
It is mainly aimed at informing people who like to trade, but provides valuable insight, into the actions of "Market Movers", "Smart Money" and "Professional Shorters".
Particularly interesting, was his statements that say, these individuals/algorithms, can see all orders in the system.
Including "stop loss" which are basically sell orders.
And the "dropping of the penny" for me, that they are responsible, for the big moves, in both directions, as traps..
These measures, apply to all markets and will be played out, on a larger scale, in the US tonight.
Incidently, I've only watched 2 or 3, of this guy's videos, a while back, but he's responsible for my adoption of Heikin-Ashi candles, when I do check the charts..
Last edited:
Food4 thought
Regular
Hey Tech where do you have a property in Nz
I have a little place on waiheke island, I love the place but never to get to spend enough time there, hopefully things will change and I will have the funds to support myself soon if all Things go the way I think we they will.
Go you good thing BRN…
Rise from the ashes
Regular
Excellent post on LinkedIn @chapman89 just had a read and a like.
DingoBorat
Slim
Pray tell..
Rise from the ashes
Regular
Not sure if you got a LinkedIn account.
www.linkedin.com
Jesse Chapman on LinkedIn: #lstm #ecosystem #ai #microwatts #cloud #chip #patent #technology…
BrainChip time has come: I think it is time to call out sceptics of the unique & unrivalled technology advance created by Peter Van Der Made & Anil Mankar…
cosors
👀
Those who are well versed in the subject can may ignore this post.
I'm responding to you because I read this and it was new to me, I've only been around a short time.
May 29, 2020
"The startup Brainchip therefore set itself the goal of developing an SNN-based chip with the Akida that is suitable for use in edge devices. It was actually supposed to be on the market last year. But then co-founder and CTO Peter van der Made realized that potential users would not be able to do much with a pure SNN chip simply because they are used to CNNs. So Brainchip decided to also integrate CNN functions on the Akida. "That's why we accepted the additional development time of one year and integrated MAC arrays on the chip," explained Peter van der Made to Markt&Technik."
I guess that's still the case, it can be chosen between SNN and CNN, just that it can't be processed as fast and energy efficient? I hope I understand this correctly. So if that's the case then I'll have to revise my original answer a bit because I would have been wrong. Please bear with me on my newbe post.
Here is a slightly older (2020) article, short and easy to understand. For those who sometimes can not follow the content and wonder what Diogenese talks about.
I don't think it's been shared here yet. It starts with Akida and ends with ReRAM from Weebit in two parts:
https://www.elektroniknet.de/halbleiter/dem-gehirn-immer-aehnlicher.176820.html
I'm not trying to add the two together, Diogenese has already made it clear that this would be difficult if I have understood this correctly. So not compatible without further ado.
Last edited:
Fullmoonfever
Top 20
Further to the earlier info on Siemens, just found a recent blog post (June2) by them as below.
No reference to BRN or Nviso obviously however definitely appear to be fans of neuromorphic AI though author thinks space may be first major start
Maybe didn't know of the Nviso collaboration
Neuromorphic systems and evolutionary AI
By Spencer Acain • June 2, 2022 • 3 MIN READ
In the field of artificial intelligence, the idea of evolution is a critical one. The very concept of an AI algorithm is one that can grow, change, adapt, and in essence evolve to fit its design requirements. AI research has progressed to such a point that an AI can be trained to achieve a level of performance far surpassing what a human could attain under specific conditions, such as a smart robotics station, offering never before seen precision and speed in smart factories.
However, this high degree of speed and precision does introduce some downsides as well. Training AI to work well in a specific location and task requires a vast amount of data from that location. What’s more, if conditions change, the algorithm may suffer reduced accuracy or cease to function altogether. Evolutionary research and neuromorphic systems is a promising approach for helping AI evolve in a more natural way, through both software and hardware, allowing it to tackle a wider range of conditions without loss of performance.
Creating a robust solution is one of the most important aspects of designing an AI system, especially for industrial applications. AI algorithms are trained to accommodate a wide expected range of conditions based on their expected operation environment, but when incoming data falls outside of the range of training data or becomes too noisy, the system will more than likely fail. Unlike AI, humans can continue to perform tasks even under widely varying conditions, so the goal is to design AI to be more human-like. This is exactly what neuromorphic systems propose to do by seeking to emulate not just the neuron structure of the brain, but the function of the synapses as well through memristors. Synapsis and memristors have the property of changing resistance based on how much current has flowed through them in the past, essentially creating a memory of past values. Thanks to this memory-like functionality the AI decision-making process becomes more robust and less susceptible to failing from less-than-ideal input data.
Going hand in hand with neuromorphic systems is evolutionary research that fundamentally is similar to reinforcement learning but does have some key differences. Reinforcement learning is a common way for AI to be trained by governing a system with a reward function that reinforces correct behaviors and punishes poor ones. This allows the software essentially on its own to evolve a solution that fits the requirements. Where an evolutionary system differs is in the scope of what the reward function manages. Called genetic code, its scope goes far beyond training just software, instead it governs everything from robot design to software to control chip layout. While in a factory the options for robot design are limited, the ability to train robotics control hardware through programmable FPGAs along with the software presents a unique path to improving AI robustness when combined with neuromorphic systems.
Allowing the hardware to be trained alongside the software under these new principles paves the way for an entirely new level of adaptability and robustness in AI. Compared to a traditional machine learning training schema, a neuromorphic system does not need to be trained with only “correct” data. Rather its training process can include non-ideal and highly variable data or even outright failure states. Although it might seem counter-productive to include undesirable outcomes in the training process, this approach allows the system to learn and evolve so it can function accurately even when contending with poor quality data or faulty hardware. More than accounting for failures, a neuromorphic system might also help mitigate the heavy reliance AI has on deployment-specific training data. Rather than resilience against poor quality data, the system would be accounting for environmental differences between the training location and the final deployment site.
The quest for truly human-like AI and robotics still has a long way to go but evolutionary design and neuromorphic systems combining to allow hardware and software to truly evolve side-by-side bring us a step closer to true natural evolution. While this technology still requires substantial development before it is economically viable to replace existing AI implementations in phones and factories, it may see its first applications in the cold reaches of space. The extreme conditions found even in low earth orbit make it difficult for existing AI systems to function accurately, something of paramount importance when operating a million miles from earth. In years to come, neuromorphic systems may become the living mind of the next generation of satellites and deep space exploration crafts pushing the boundary of human understanding.
Siemens Digital Industries Software is driving transformation to enable a digital enterprise where engineering, manufacturing and electronics design meet tomorrow. Xcelerator, the comprehensive and integrated portfolio of software and services from Siemens Digital Industries Software, helps companies of all sizes create and leverage a comprehensive digital twin that provides organizations with new insights, opportunities and levels of automation to drive innovation.
For more information on Siemens Digital Industries Software products and services, visit siemens.com/software or follow us on LinkedIn, Twitter, Facebook and Instagram.
Siemens Digital Industries Software – Where today meets tomorrow.
No reference to BRN or Nviso obviously however definitely appear to be fans of neuromorphic AI though author thinks space may be first major start
Maybe didn't know of the Nviso collaboration
Neuromorphic systems and evolutionary AI - Thought Leadership
In the field of artificial intelligence, the idea of evolution is a critical one. The very concept of an AI algorithm is one that can grow, change, adapt, and
blogs.sw.siemens.com
Neuromorphic systems and evolutionary AI
By Spencer Acain • June 2, 2022 • 3 MIN READ
In the field of artificial intelligence, the idea of evolution is a critical one. The very concept of an AI algorithm is one that can grow, change, adapt, and in essence evolve to fit its design requirements. AI research has progressed to such a point that an AI can be trained to achieve a level of performance far surpassing what a human could attain under specific conditions, such as a smart robotics station, offering never before seen precision and speed in smart factories.
However, this high degree of speed and precision does introduce some downsides as well. Training AI to work well in a specific location and task requires a vast amount of data from that location. What’s more, if conditions change, the algorithm may suffer reduced accuracy or cease to function altogether. Evolutionary research and neuromorphic systems is a promising approach for helping AI evolve in a more natural way, through both software and hardware, allowing it to tackle a wider range of conditions without loss of performance.
Creating a robust solution is one of the most important aspects of designing an AI system, especially for industrial applications. AI algorithms are trained to accommodate a wide expected range of conditions based on their expected operation environment, but when incoming data falls outside of the range of training data or becomes too noisy, the system will more than likely fail. Unlike AI, humans can continue to perform tasks even under widely varying conditions, so the goal is to design AI to be more human-like. This is exactly what neuromorphic systems propose to do by seeking to emulate not just the neuron structure of the brain, but the function of the synapses as well through memristors. Synapsis and memristors have the property of changing resistance based on how much current has flowed through them in the past, essentially creating a memory of past values. Thanks to this memory-like functionality the AI decision-making process becomes more robust and less susceptible to failing from less-than-ideal input data.
Going hand in hand with neuromorphic systems is evolutionary research that fundamentally is similar to reinforcement learning but does have some key differences. Reinforcement learning is a common way for AI to be trained by governing a system with a reward function that reinforces correct behaviors and punishes poor ones. This allows the software essentially on its own to evolve a solution that fits the requirements. Where an evolutionary system differs is in the scope of what the reward function manages. Called genetic code, its scope goes far beyond training just software, instead it governs everything from robot design to software to control chip layout. While in a factory the options for robot design are limited, the ability to train robotics control hardware through programmable FPGAs along with the software presents a unique path to improving AI robustness when combined with neuromorphic systems.
Allowing the hardware to be trained alongside the software under these new principles paves the way for an entirely new level of adaptability and robustness in AI. Compared to a traditional machine learning training schema, a neuromorphic system does not need to be trained with only “correct” data. Rather its training process can include non-ideal and highly variable data or even outright failure states. Although it might seem counter-productive to include undesirable outcomes in the training process, this approach allows the system to learn and evolve so it can function accurately even when contending with poor quality data or faulty hardware. More than accounting for failures, a neuromorphic system might also help mitigate the heavy reliance AI has on deployment-specific training data. Rather than resilience against poor quality data, the system would be accounting for environmental differences between the training location and the final deployment site.
The quest for truly human-like AI and robotics still has a long way to go but evolutionary design and neuromorphic systems combining to allow hardware and software to truly evolve side-by-side bring us a step closer to true natural evolution. While this technology still requires substantial development before it is economically viable to replace existing AI implementations in phones and factories, it may see its first applications in the cold reaches of space. The extreme conditions found even in low earth orbit make it difficult for existing AI systems to function accurately, something of paramount importance when operating a million miles from earth. In years to come, neuromorphic systems may become the living mind of the next generation of satellites and deep space exploration crafts pushing the boundary of human understanding.
Siemens Digital Industries Software is driving transformation to enable a digital enterprise where engineering, manufacturing and electronics design meet tomorrow. Xcelerator, the comprehensive and integrated portfolio of software and services from Siemens Digital Industries Software, helps companies of all sizes create and leverage a comprehensive digital twin that provides organizations with new insights, opportunities and levels of automation to drive innovation.
For more information on Siemens Digital Industries Software products and services, visit siemens.com/software or follow us on LinkedIn, Twitter, Facebook and Instagram.
Siemens Digital Industries Software – Where today meets tomorrow.
DingoBorat
Slim
"I guess that's still the case, it can be chosen between SNN and CNN, just that it can't be processed as fast and energy efficient?"
My understanding Cosors, is that AKIDA takes the Convolurional Neural Network and Spikenises it and so in doing so, retains the lion's share of the Spiking NNs benefits
Rise from the ashes
Regular
Good morning @Esq.111
Yes I found it a very insightful read
cheers for taking the time to have a read. Glad you found it worth your while.
Last edited:
Thanks for this cosors,
As you say, it explains the concepts in readily-understandable language.
Although I had deduced the need for 2-bit and 4-bit weights/actuations Akida to use MAC (Multiply Accumulate) calculation circuits, this is one of the first publications I have seen which attributes the presence of MACs to a statement from the company.
There is a trade off between speed/power efficiency and accuracy as the number of bits in the weights/actuations increases.
A 4*4 MAC is roughly speaking 4 times faster/less power hungry than an 8*8 MAC.
A 1-bit (pure spiking) Akida is about 16 times faster/more power efficient than the 4*4 MAC Akida embodiment. [edit: However different layers in Akida can have different numbers of bits, so for example, a first layer may use 4-bits and the next 2 layers may use 1 bit, and the output may (possibly?) use 4 bits. ]
As the article you attached says, Weebit are still trying to get a consistently reproduceable analog MemRistor, as the manufacturing process is difficult to control precisely. Manufacturing variations are a problem for analog neurons/synapses because there are hundreds or thousands of input signals (currents) for each synapse in which the input currents are added to produce an output voltage whose amplitude is determined by the sum of the input currents flowing in a resistor. That is, for each input current, the output voltage increases by a fixed amount. Hence the operating voltage of the circuit must be divided into very small increments to accommodate the number of potential input currents. Thus variations in the resistance of the MemRistor/ReRAM elements can produce errors which accumulate.
The reason this is not a problem with digital neurons/synapses is that the digital voltage has a much greater margin for error because the operating voltage only needs to be divided in half to indicate either a 1 or a zero, and the accumulation of signals is a digital number composed of 1s or zeros.
Last edited:
On the SNN/CNN issue, CNN is a software technique running MACs on CPU/GPU, hence slow and power hungry.
CNN can be converted to run on Akida SoC using the Akida CNN2SNN toolkit, but it isn't quite plug-and-play.
https://doc.brainchipinc.com/user_guide/cnn2snn.html
The Brainchip CNN2SNN toolkit provides means to convert Convolutional Neural Networks (CNN) that were trained using Deep Learning methods to a low-latency and low-power Spiking Neural Network (SNN) for use with the Akida runtime. This document is a guide to that process.
The Akida neuromorphic IP provides Spiking Neural Networks (SNN) in which communications between neurons take the form of “spikes” or impulses that are generated when a neuron exceeds a threshold level of activation. Neurons that do not cross the threshold generate no output and contribute no further computational cost downstream. This feature is key to Akida hardware efficiency. The Akida neuromorphic IP further extends this efficiency by operating with low bitwidth “synapses” or weights of connections between neurons.
Despite the apparent fundamental differences between SNNs and CNNs, the underlying mathematical operations performed by each may be rendered identical. Consequently, the trained parameters of a CNN can be converted to be Akida-compatible, given only a small number of constraints. By careful attention to specifics in the architecture and training of the CNN, an overly complex conversion step from CNN to SNN can be avoided. The CNN2SNN toolkit comprises a set of functions designed for the popular Tensorflow Keras framework, making it easy to train a SNN-compatible network.
cosors
👀
Thank you very much for both contributions!
Even I understand that.
I'm glad you were able to find something in the article as well (hint regarding MACs). Here is another report that I have decided to read and that was mentioned here some time ago. Only he does not resolve the matter as accurately as you with the other post and the CNN2SNN toolkit. This is probably one of the reports that cost a lot of money initially and perhaps focuses more on research regarding Brainchip for investment..
"Akida can perform event-based convolution as well, enabling the chip to run both Convolutional Neural Networks (CNN, see Appendix I) and Spiking Neural Networks (SNN). BrainChip is the only company to have combined neuromorphic processing with event-based convolution."
Welcome to the revolution
https://static1.squarespace.com/sta...s+research+initiation+report+-+20+08+2021.pdf
Similar threads
