BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Fact Finder
Top 20
Blind Freddie said I had to give a heart for this post. LOLI am sure this has been noted already, but here it is again. We know that Information Systems Laboratories was a BrainChip Early Access Partner and we know that they are collaborating with BrainChip on the development of a radar system for the Air Force Research Laboratory. If you take a close look at ISL's website you will find the following:
Neuromorphic Engineering and Artificial Intelligence
ISL is focused on replicating the analog nature of biological computation and the role of neurons in cognition. ISL’s team of scientists/engineers continue to understand how the morphology of individual neurons, circuits, applications, and overall architectures creates desirable computations. Leveraging this understanding and the newly developed and emerging commercial neuromorphic chips, ISL is developing a new low-power, lightweight detect and avoid (DAA) system for very small UAS platforms that exploits automotive radar hardware, light-weight EO/IR sensors, advanced data fusion algorithms, and neuromorphic computing.
This is just another example of a really smart company that is successful in the technological field, that does its research, and that has chosen BrainChip's Akida as the product that it needs. ISL, Renesas, MegaChips, NASA, Valeo, Vorago........how much more obvious does it have to become before others wake up to what blind Freddy can see, that BrainChip is on a path to historical success and dominance in the field of artificial intelligence.
Really is special being a part of the BrainChip community on TSex. We know what much of the World does not.
On a different note I have grabbed out the following paragraph:
“very small UAS platforms that exploits automotive radar hardware, light-weight EO/IR sensors, advanced data fusion algorithms, and neuromorphic computing.”
ISL is taking automotive radar hardware that is already in the market and making it smart in accordance with the Brainchip mantra - we don’t make sensors we make them smart - and if as it seems on a balance of probabilities that they do take this automotive radar and make it smart enough for the US Airforce it seems reasonable that it would be immediately marketable in the automotive industry.
The sales pitch: “You know that dumb as bat droppings radar sensor, we have the same one but it’s now smart and uses hardly any power, and reduces overall system energy cost by 5 to 10 times over the competition.” “Is it any good?” “Mate it’s used by the US Airforce. It’s using the smarts that NASA and Mercedes use. It’s hot as. But you can have the dumb one if you want.”
This kind of kicks mud in the faces of those over at that other place who constantly denied the financial value of these NASA and DARPA relationships.
My opinion only DYOR
FF
AKIDA BALLISTA
Bravo
If ARM was an arm, BRN would be its biceps💪!
Maybe Intel just broke cover? They may have decided it's too hard to get up to speed with BrainChip? Like, if you can't beat them, join them?
Last edited:
it seems that Trey is toeing the party line in implying that Loihi is state-of-the-art in neuromorphic computing. - Dio
---------------------------------------------------------------------------------------------
Yes it would seem its a case of ALL THINGS - INTEL. (only)
In the early hours of this morning (Fri) I was up watching some CNBC and they had an interview between Cramer & another person. It was all about INTEL and its position in the world of (IoT).
Cramer was stating the obvious that INTELs best days are behind it and that it is trading on its name only. The "other" person was a fanatic "all things are INTEL" person which made it very interesting to watch.
Crammers position was that even with Government (sponsership) (ie: Grants), that INTELS net Profit ($19.8 Bil) was not enough to venture into the world of FABs and how was it going to finance all this especially when INTEL is sliding behind in the CHIP world being quickly outpaced by others in this field.
(BRN came to mind here! Go AKIDA. lol)
We have known for some time (several years) how INTEL has not succeeded in the Neuromorphic Chip arena with anything more than a research chip with limited abilities. We have talked about how INTEL has fallen behind regardless of its R&D budget and sizable asset base!
It would SEEM that many others have begun to notice this fact also and that by just invoking the name INTEL does not mean the product is the market leader or even in the running now days! About time for this recognition.
To many IoT people and industry "experts" have treated INTEL as a god and only bother to use them as case reference material instead of doing some real research and discovering other companies and products (like Brainchip) are out there and have left INTEL behind in their dust!
Crammer was actually suggesting that INTEL and also Pat were doing a lack lustre job at best and there were stormy seas in front of both in the near future! Being the "rescuer" of the USA for Chip production was not going to be easy nor guaranteed it would seem.
Interesting world we live in. How the big can fall.
Yak52
ps. I have always had INTEL as CPU in my computers but next time will be looking elsewhere.
My thinking is that the current owners, should they want to sell it off as they seem to be saying, would want to dress it up as a Tesla alternative. Sadly, they wont have the time to develop their own AV setup and should be forced to grab what there is currently out there that can give them some cachet. Who else could they turn to?
TechGirl
Founding Member
"Ultimately, the tech could provide the basis for a system that improves traffic flow for all road users"
Wouldn't that be nice
Traffic lights that actually control "traffic"..
It's only 2022, maybe I expect too much...
Hey DB,
It would be wonderful, I was waiting at the lights just the other day, absolutely no cars going through and I was thinking come on world hurry up and implement Brainchip, I had an important date with a rum and coke and the dumb lights were holding me up
Hi Alfie,
A term from mining shares - nearology.
HRL's zero-shot learning is software-intensive ( WO2021118697A1, WO2020159638).
HRL are long on memristors (US11238335B2).
WO2021118697A1 PROCESS TO LEARN NEW IMAGE CLASSES WITHOUT LABELS
Described is a system for learning object labels for control of an autonomous platform. Pseudo-task optimization is performed to identify an optimal pseudo-task for each source model of one or more source models. An initial target network is trained using the optimal pseudo-task. Source image components are extracted from source models, and an attribute dictionary of attributes is generated from the source image components. Using zero-shot attribution distillation, the unlabeled target data is aligned with the source models similar to the unlabeled target data. The unlabeled target data are mapped onto attributes in the attribute dictionary. A new target network is generated from the mapping, and the new target network is used to assign an object label to an object in the unlabeled target data. The autonomous platform is controlled based on the object label.
WO2020159638A1 SYSTEM AND METHOD FOR UNSUPERVISED DOMAIN ADAPTATION VIA SLICED-WASSERSTEIN DISTANCE
Described is a system for unsupervised domain adaptation in an autonomous learning agent. The system adapts a learned model with a set of unlabeled data from a target domain, resulting in an adapted model. The learned model was previously trained to perform a task using a set of labeled data from a source domain. The set of labeled data has a first input data distribution, and the set of unlabeled target data has a second input data distribution that is distinct from the first input data distribution. The adapted model is implemented in the autonomous learning agent, causing the autonomous learning agent to perform the task in the target domain.
US11238335B2 Active memristor based spiking neuromorphic circuit for motion detection
A motion-sensing circuit for determining a direction of motion and a velocity of an object includes a first photo-receptor for sensing the object, an excitatory active memristor neuron circuit coupled to the first photo-receptor, a second photo-receptor for sensing the object, an inhibitory active memristor neuron circuit coupled to the second photo-receptor, and a self-excitatory active memristor output-counter neuron circuit coupled to the excitatory active memristor neuron circuit and coupled to the inhibitory active memristor neuron circuit.
Silly girl,Hey DB,
It would be wonderful, I was waiting at the lights just the other day, absolutely no cars going through and I was thinking come on world hurry up and implement Brainchip, I had an important date with a rum and coke and the dumb lights were holding me up
That's why nostrils come in pairs.
TechGirl
Founding Member
Dallas
Regular
Rugged computing takes on the battlefield networking challenge
Connectivity, networking, SWaP, and thermal management are among the highest priorities of rugged computing designers to support mobile warfighters at the edge.
TechGirl
Founding Member
Yes there’s lots of stupid governments around the globe hopefully akida will make them smarterOther countries aren't suffering from the lack of ai from what I hear... just stupid government people. I think they like frustrating people. They don't want happy people..
RobjHunt
Regular
Hey TT77, certainly not speculating. Yep, the quarterlies. Due to the NDA’s & EAP’s & ABC’s an AEIOU’s (sorry, might have gone too far there
) it’s my opinion that then is an assumption of an assured payday I may very well be wrong but I don’t think so
Pantene!
Bravo
If ARM was an arm, BRN would be its biceps💪!
OK? It's late and I don't really get it. Why do balls come in pairs then?
JK200SX
Regular
Dunno, but here's another one...... Why can't gypsies have babies?
My thinking is they are mining very close to AKIDA califirnication territory
You don't just step down from one to none without knowing something, plus there's 100 NDASand I'm liking my dot join to HRL labs..afterall PVDM cited their spiking neural networks patent from 2012 in 2025 from memory
The concept of zero-shot learning has been around for a few years as explained in the HRL patent I cited. In this document, the zero-shot learning process is software intensive carried out on a programmable computer.
WO2021118697A1 PROCESS TO LEARN NEW IMAGE CLASSES WITHOUT LABELS
[00043] The following references are cited and incorporated throughout this application. For clarity and convenience, the references are listed herein as a central resource for the reader. The following references are hereby incorporated by reference as though fully set forth herein. The references are cited in the application by referring to the corresponding literature reference number, as follows: 1. Xian, Y., Schiele, B. and Akata, Z., 2017. Zero-shot learning-the good, the bad and the ugly. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.4582-4591 ...
[00067] (3.5) Step 5: Zero-Shot Attribute Distillation (element 508)
[00068] Given the attributes from the sources in the attribute dictionary (element 540), the initial target network (element 526), and the source models, the unlabeled target data (element 522) needs to be processed so that the unlabeled target data (element 522) maps onto the right set of attributes (i.e., map target data onto abstract attributes (element 542). To achieve this mapping, the closest source model or models is selected based on the SSG (element 516) and then a zero-shot attribute distillation method (element 544) is used to align the sources with the target. Alignment happens by learning a shared latent space for the source and target models that is predictive of the attributes for the source data and simultaneously generative of the target data. As a result, there is a new target network (element 546) that maps target data onto a probability distribution over the abstract attributes in the attribute dictionary (element 540). This distribution can then be mapped onto a label or sentence. Note that in all of these five steps, not a single training label was used to assign labels to data from a new domain.
Claim 1. A system for learning object labels for control of an autonomous platform, the system comprising:
one or more processors and a non-transitory computer-readable medium having executable instructions encoded thereon such that when executed, the one or more processors perform operations of:
performing pseudo-task optimization to identify an optimal pseudo-task for each source model of one or more source models;
training an initial target network with self-supervised learning using the optimal pseudo-task;
extracting a plurality of source image components from the one or more source models;
generating an attribute dictionary of abstract attributes from the plurality of source image components;
using zero-shot attribution distillation, aligning a set of unlabeled target data with the one or more source models that are similar to the set of unlabeled target data;
mapping the set of unlabeled target data onto a plurality of abstract attributes in the attribute dictionary; generating a new target network from the mapping;
using the new target network, assigning an object label to an object in the unlabeled target data; and controlling the autonomous platform based on the assigned object label.
Similar threads
