BRN Discussion Ongoing

Xray1

Regular
  • Like
  • Fire
  • Love
Reactions: 34 users

butcherano

Regular
Well they do not appear on their list of partners. If Brainchip fills the gap on the list over the next week or so then I suspect they have moved on.

Perhaps SynSense web site will tell the story.

My opinion only DYOR
FF

AKIDA BALLISTA
Yeah…interesting that SynSense aren’t listed as a partner on the About tab here…https://www.prophesee.ai/about-prophesee/

But they’re still listed as a partner on the homepage here…. https://www.prophesee.ai/

Will be interesting to see if they’re still there when Brainchip make an appearance on the list…(y)
 
  • Like
  • Fire
  • Love
Reactions: 27 users

Reuben

Founding Member
  • Like
  • Fire
  • Love
Reactions: 34 users

Xray1

Regular
This looks better :)

View attachment 9679
DYOR, I've saved the hard work for Prophesee, by showing how nice their page would look like with our company logo.
I think our Co name would look a lot better if it was placed at the "top" of the partners lists.... lol
 
  • Like
  • Fire
  • Love
Reactions: 17 users

butcherano

Regular
  • Like
  • Fire
  • Thinking
Reactions: 28 users
"We've successfully ported the data from Prophesee's neuromorphic-based camera sensor to process inference on Akida with impressive performance," said Anil Mankar, Co-Founder and CDO of BrainChip. "This combination of intelligent vision sensors with Akida's ability to process data with unparalleled efficiency, precision and economy of energy at the point of acquisition truly advances state-of-the-art AI enablement and offers manufacturers a ready-to-implement solution."

"By combining our Metavision solution with Akida-based IP, we are better able to deliver a complete high-performance and ultra-low power solution to OEMs looking to leverage edge-based visual technologies as part of their product offerings, said Luca Verre, CEO and co-founder of Prophesee."

Its great to be a shareholder
If you take the following part of Prophesee’s statement above:

“"By combining our Metavision solution with Akida-based IP, we are better able to deliver a complete high-performance and ultra-low power solution to OEMs”

with what they said when they partnered with SynSense in October, 2021 it is clear that SynSense was not all they thought it was going to be:



If you were SynSense you would cringe at the statement that Brainchip is ‘better able’ to do what they were supposed to in the above announcement.

Ubiquitous. DEFAULT STANDARD.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Love
  • Fire
Reactions: 50 users

wilzy123

Founding Member
If you take the following part of Prophesee’s statement above:

“"By combining our Metavision solution with Akida-based IP, we are better able to deliver a complete high-performance and ultra-low power solution to OEMs”

with what they said when they partnered with SynSense in October, 2021 it is clear that SynSense was not all they thought it was going to be:



If you were SynSense you would cringe at the statement that Brainchip is ‘better able’ to do what they were supposed to in the above announcement.

Ubiquitous. DEFAULT STANDARD.

My opinion only DYOR
FF

AKIDA BALLISTA
WASCA - What's a SynSense Chip Anyway
 
  • Haha
  • Like
Reactions: 37 users
Therein lies the weakness of your argument ‘as far as we know’.

Given we all know we do not know everything that is taking place behind the Brainchip veil your certainty is certainly misplaced.

If you read what Anil Mankar said at 2021 Ai Field Day he left wide open the opportunity that LSTM could be added to AKD1000 neural fabric by software without changes to the fabric as it already uses ‘itegrate and fire neuron’.

So put your briefs back on for the moment my speculation is still taking up the surface of the table.

My opinion only DYOR
FF

AKIDA BALLISTA
Hi FF,

Personally I disagree with all the comments said here about this. The word inference has different meanings depending on which industry you are referring to. In the context of machine learning, inference refers to the data being fed into the machine learning model or algorithm (see link below).

So in the context here, processing inference simply refers to processing sensor data. Nothing further to read here, definitely no reference to LSTM's.
On top of this, since English isn't Anil's primary language, I think some leniency may be required, so there's not much point reading too deeply into how his sentences are structured.

Pure speculation, DYOR


Machine learning inference is the process of running data points into a machine learning model to calculate an output such as a single numerical score. This process is also referred to as “operationalizing a machine learning model” or “putting a machine learning model into production.”
 
  • Like
  • Love
  • Fire
Reactions: 30 users

LuWil

Regular
1A32D054-D5BD-49CB-8A39-AC011A846518.jpeg


9B35EF77-E988-4E24-9D40-159EEFBC2F38.jpeg


0A725E17-5472-4FA7-9F96-021E2167F7C4.jpeg


251ED420-6BBA-4BD2-9E94-F625DC82DCE9.jpeg
 
  • Like
  • Fire
  • Love
Reactions: 34 users
The following is dedicated to the folks at SynSense:

 
  • Haha
  • Like
Reactions: 21 users
Hi FF,

Personally I disagree with all the comments said here about this. The word inference has different meanings depending on which industry you are referring to. In the context of machine learning, inference refers to the data being fed into the machine learning model or algorithm (see link below).

So in the context here, processing inference simply refers to processing sensor data. Nothing further to read here, definitely no reference to LSTM's.
On top of this, since English isn't Anil's primary language, I think some leniency may be required, so there's not much point reading too deeply into how his sentences are structured.

Pure speculation, DYOR


Machine learning inference is the process of running data points into a machine learning model to calculate an output such as a single numerical score. This process is also referred to as “operationalizing a machine learning model” or “putting a machine learning model into production.”
I am still speculating. The logic is that the next generation with LSTM has been available to early access customers and robotics needs LSTM and Prophesee have jumped ship mid voyage.

I am joining the @MC🐠 school of early speculation.

Prove that Brainchip has not released LSTM to early access customers as it did with AKD1000 IP before announcing its availability to the market generally and I will give up speculating.

We both know you cannot anymore than I can prove it has been so I will continue to rely on the history of how Brainchip has done things in the past and the unusual language.

My opinion and speculation only so DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Love
  • Haha
Reactions: 34 users

Proga

Regular
  • Like
  • Fire
  • Love
Reactions: 36 users

LuWil

Regular

Researchers at the Collaborative, Learning, and Adaptive Robots (CLeAR) Lab and TEE Research Group at National University of Singapore are taking advantage of the benefits of Event-Based Vision, in combination with touch, to build new visual-tactile datasets for the development of better learning systemsin robotics. The neuromorphic sensor fusion of touch and vision is being used to help robots grip and identify objects.

Their research work on ‘electronic skin’ includes Prophesee’s Metavision Event-Based Vision sensor, and demonstrates how touch and vision can be combined for enhanced capabilities in robotics.

78C329F8-136F-48E3-B44B-AC2B77057095.jpeg


This looks like it all ties in with TACNIQ which was founded by Benjamin Tee (TEE Research Group) which as we know, are currently evaluating Akida.


My opinion only.
DYOR
 
  • Like
  • Fire
  • Love
Reactions: 29 users

Diogenese

Top 20
Hi FF,

Personally I disagree with all the comments said here about this. The word inference has different meanings depending on which industry you are referring to. In the context of machine learning, inference refers to the data being fed into the machine learning model or algorithm (see link below).

So in the context here, processing inference simply refers to processing sensor data. Nothing further to read here, definitely no reference to LSTM's.
On top of this, since English isn't Anil's primary language, I think some leniency may be required, so there's not much point reading too deeply into how his sentences are structured.

Pure speculation, DYOR


Machine learning inference is the process of running data points into a machine learning model to calculate an output such as a single numerical score. This process is also referred to as “operationalizing a machine learning model” or “putting a machine learning model into production.”
Hi IDD,

You've left out the outference "to calculate an output".

The inference is the estimating of the output from the input data.

infer
[ɪnˈfəː]
VERB

  1. deduce or conclude (something) from evidence and reasoning rather than from explicit statements:


Inference, from my understanding, is the classification of an object from data which is similar to, but not identical with already known data.

Akida has been processing inference for donkeyears, as exemplified by the use of the word 11 times in this 2019 presentation.

https://brainchip.com/wp-content/uploads/2019/10/BrainChip-Linley-Akida-Presentation_v5.pdf


For example, Akida may determine the classification of an object if it has a sufficient percentage of similarities to a known object, or if a part of the new object is sufficiently similar to a corresponding part of a known object.
 
  • Like
  • Love
  • Fire
Reactions: 16 users

uiux

Regular
For what it's worth:


Python:
def infer(self):
        while True:
            input_array = camera.get_input_array()
            predictions = self.model_ak.predict_classes(input_array, num_classes=NUM_CLASSES)
            if predictions[0] in SAVED:
                self.camera.label = LABELS.get(predictions[0], predictions[0])
                self.camera.shots = "{} shot/s".format(SHOTS.get(predictions[0]))
            time.sleep(1 / INFERENCE_PER_SECOND)


This is logic to get a frame from a webcam to process with akida, eg. the one shot learning demo
 
  • Like
  • Fire
  • Love
Reactions: 34 users
Anyone else remember what Ryad Benosman the founder of Prophesee said on social media when asked by a poster here about Brainchip and AKIDA.

Be interesting to have it reposted here in light of what has been announced today.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Thinking
  • Fire
Reactions: 20 users

Foxdog

Regular
Still no major pop on the SP - is the market blind 😆
 
  • Like
  • Thinking
Reactions: 8 users

Mccabe84

Regular
  • Like
  • Love
  • Thinking
Reactions: 12 users

M_C

Founding Member
I was saving these for a rainy day............

Capture.PNG
Capture1.PNG
Capture2.PNG
 
  • Like
  • Love
  • Fire
Reactions: 79 users

Lex555

Regular
Playing devils advocate for a minute, my calculate revenue by royalties x price of royalties. With all these partnerships coming out and interconnecting with their own ecosystems it would be easy to imagine market(s) saturation. @MC🐠 thanks.

What I’m not 100% convinced on is average price per royalty, could we be potentially over estimating here? @Fact Finder what do you think?

Comparing ARM Holdings, they have 2021 revenue of $2.7b from a total of ~$30b chips sold this year which gives an average of $0.10 revenue per chip. Why is this so low?


Makes me wonder if we’re doing IP, and agreements aren’t disclosed, then what are the chances we could be overstating this 10x.

Or does it even matter, with a conservative $60b market by 2025 and Akida being best in class/ first in class we achieve a more likely 50% share like @chapman89 says, then multiply by 15x (ARM ranges from 10-25x) we get $450bn USD MC. Hmm…
 
  • Like
  • Thinking
  • Love
Reactions: 27 users
Top Bottom