BRN Discussion Ongoing

[Learning]

This is intresting, I wonder which Neuromorphic Computing/ Spiking Neural Network (SNNSs) was used?.
After last night post from thelitteshort, it's most probably Brainchip's Akida. As their are no neuromorphic chips commercially available, other than Brainchip.

There's an interesting story behind this paper. | Arijit Mukherjee

There's an interesting story behind this paper. That was a time when we were presenting our findings on neuromorphic at a workshop attended by many researchers from Tata Consultancy Services - Research. We stressed on the capability of SNNs that allow them to learn quickly from less data. The...

www.linkedin.com

Spiking Neural Networks, coupled with Neuromorphic computing platforms, promises low-power training and inference at battery driven Edge devices. | Sounak Dey

Spiking Neural Networks, coupled with Neuromorphic computing platforms, promises low-power training and inference at battery driven Edge devices. In a recent paper titled "An SNN Based ECG Classifier For Wearable Edge Devices", accepted at NeurIPS 2022 TS4H workshop, and co-authored by...

www.linkedin.com

Learning

 

[Rise from the ashes]

This is intresting, I wonder which Neuromorphic Computing/ Spiking Neural Network (SNNSs) was used?.
After last night post from thelitteshort, it's most probably Brainchip's Akida. As their are no neuromorphic chips commercially available, other than Brainchip.

View attachment 23171 View attachment 23174
View attachment 23173

There's an interesting story behind this paper. | Arijit Mukherjee

There's an interesting story behind this paper. That was a time when we were presenting our findings on neuromorphic at a workshop attended by many researchers from Tata Consultancy Services - Research. We stressed on the capability of SNNs that allow them to learn quickly from less data. The...

www.linkedin.com

Spiking Neural Networks, coupled with Neuromorphic computing platforms, promises low-power training and inference at battery driven Edge devices. | Sounak Dey

Spiking Neural Networks, coupled with Neuromorphic computing platforms, promises low-power training and inference at battery driven Edge devices. In a recent paper titled "An SNN Based ECG Classifier For Wearable Edge Devices", accepted at NeurIPS 2022 TS4H workshop, and co-authored by...

www.linkedin.com

Learning

Been walking down memory Lane again and tripped over this.

When Robot?

 

[Fact Finder]

Been walking down memory Lane again and tripped over this.
View attachment 23175
When Robot?

I have posted about this since reading the address of the CEO of Tata to shareholders back in 2020.

In short he was apologising to shareholders for the need to maintain strict secrecy around their projects and customers until product release because of Chinese sponsored industrial espionage.

After the Brainchip and Tata Consulting Services joint presentation on 14.12.19 showing AKIDA performing live gesture identification of hand signals from audience members things went completely quiet.

The usual crowd used this silence to claim the relationship had ended in failure despite the clear statement by Tata’s CEO and despite the scientific papers being published by TCS members.

While it is speculation on my part the fact that Dr. Arijit Mukherjee and colleagues are being more outspoken could signal that they are close to releasing something involving AKIDA perhaps even his predicted robot.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Foxdog]

Anyone know off the top of their head when Akida 2000 is supposed to be released? I thought it was soon.

Tomorrow

 

[Rise from the ashes]

I have posted about this since reading the address of the CEO of Tata to shareholders back in 2020.

In short he was apologising to shareholders for the need to maintain strict secrecy around their projects and customers until product release because of Chinese sponsored industrial espionage.

After the Brainchip and Tata Consulting Services joint presentation on 14.12.19 showing AKIDA performing live gesture identification of hand signals from audience members things went completely quiet.

The usual crowd used this silence to claim the relationship had ended in failure despite the clear statement by Tata’s CEO and despite the scientific papers being published by TCS members.

While it is speculation on my part the fact that Dr. Arijit Mukherjee and colleagues are being more outspoken could signal that they are close to releasing something involving AKIDA perhaps even his predicted robot.

My opinion only DYOR
FF

AKIDA BALLISTA

U.S. Patent Application for SYSTEM AND METHOD FOR REAL-TIME RADAR-BASED ACTION RECOGNITION USING SPIKING NEURAL NETWORK(SNN) Patent Application (Application #20210365778 issued November 25, 2021) - Justia Patents Search

This disclosure relates generally to action recognition and more particularly to system and method for real-time radar-based action recognition. The classical machine learning techniques used for learning and inferring human actions from radar images are compute intensive, and require volumes of...

patents.justia.com

Yes I recall that discussion imo they are definitely still working with us.

I'm terrible at snipping so maybe someone could pick out the juicy parts of the patent

 

[Rise from the ashes]

Tomorrow

Oh good I'll save five bucks then

 

[Foxdog]

Indeed, I think you stand to make more if the announcement drops. It'll have to have a significant positive impact on the SP. Personally I reckon it should be similar to the MB announcement although imo it's of far greater significance.

 

[Fact Finder]

Tomorrow

The truth is tomorrow will never come for AKD 2000 as the AKD 1000, 2000, 3000, 4000 & 5000 series was cancelled everyone should have received a memo.

The AKIDA 1.0 replaced AKD 1000 however it could well be that there will be a further reimagining of the brand for the LSTM Edition IP.

I personally would expect that as the CEO Sean Hehir made a price sensitive announcement as follows at the AGM:

“For the short term, you can expect another major release of next generation IP and another reference chip prior to the next AGM.”

and the half yearly report accounted for monies paid for IP to produce the final engineering for the reference chip things are still moving ahead as planned.

So it is a question what does in the near term mean for the IP? At the very least it is before next years AGM and before the reference chip so my bet is before Christmas.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Foxdog]

AKIDA NEXTGEN release tomorrow then

 

[Fact Finder]

U.S. Patent Application for SYSTEM AND METHOD FOR REAL-TIME RADAR-BASED ACTION RECOGNITION USING SPIKING NEURAL NETWORK(SNN) Patent Application (Application #20210365778 issued November 25, 2021) - Justia Patents Search

This disclosure relates generally to action recognition and more particularly to system and method for real-time radar-based action recognition. The classical machine learning techniques used for learning and inferring human actions from radar images are compute intensive, and require volumes of...

patents.justia.com

Yes I recall that discussion imo they are definitely still working with us. View attachment 23183
I'm terrible at snipping so maybe someone could pick out the juicy parts of the patent

I have never doubted. It was either that or Dr. Arijit Mukherjee constantly updating his LinkedIn with all things Brainchip was because he wanted a job and a house with a view of the Harbour and Opera House in Sydney.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Rise from the ashes]

I have never doubted. It was either that or Dr. Arijit Mukherjee constantly updating his LinkedIn with all things Brainchip was because he wanted a job and a house with a view of the Harbour and Opera House in Sydney.

My opinion only DYOR
FF

AKIDA BALLISTA

Yep damn obvious things are a goer with Tata. I'm just waiting for Dr. Arijit Mukherjee to update his profile pic showing off his new Akida tattoo for further confirmation

 

[Fact Finder]

Yep damn obvious things are a goer with Tata. I'm just waiting for Dr. Arijit Mukherjee to update his profile pic showing off his new Akida tattoo for further confirmation

The other influencing factor is that Dr. Arpan Pal Head of the Robotics group and to whom Dr.Arijit Mukherjee reports directly publishes papers one of which if memory serves was titled ‘Monitoring the Health of Man and Machine’ and all of this series of the proposals/solutions would be enhanced by SNN produced by a low powered semiconductor capable of working autonomously without connection.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Rise from the ashes]

The other influencing factor is that Dr. Arpan Pal Head of the Robotics group and to whom Dr.Arijit Mukherjee reports directly publishes papers one of which if memory serves was titled ‘Monitoring the Health of Man and Machine’ and all of this series of the proposals/solutions would be enhanced by SNN produced by a low powered semiconductor capable of working autonomously without connection.

My opinion only DYOR
FF

AKIDA BALLISTA

Is this the one you are talking about?

 

[Rise from the ashes]

While we're here let's boost his numbers

 

[Fact Finder]

Whatever you can imagine, Akida can do it!

My first thought was a WANCA re-education chip but even at AKIDA’s low power it would probably be too powerful and fry their remaining synapse.

So back to the drawing board.

 

[Fact Finder]

Is this the one you are talking about? View attachment 23184

Yes but there was/is a series of papers on this point Dr. Arpan Pal has a special interest in remote health monitoring particularly for cardiac health and also aftermarket implementation for monitoring machines using vision and Neuromorphic computing (SNN).

Anyway he has delivered lectures and written numerous papers with similar titles to the one you have located.
Regards
FF

AKIDA BALLISTA

 

[Fact Finder]

The following link takes you to Dr. Arpan Pal’s many, many Google citations:

Dr. Arpan Pal

Distinguished Chief Scientist and Research Area Head, Embedded Systems, TCS Research, TCS - Cited by 6,390 - Embedded Systems - Computational Sensing - Signal/Image Processing and ML - Edge Computing and Edge AI

scholar.google.com

This is but one of dozens:

System and method for real-time radar-based action recognition using spiking neural network (snn)
Inventors
DEY Sounak, Arijit Mukherjee, Dighanchal Banerjee, RANI Smriti, Arun George, Tapas Chakravarty, Arijit Chowdhury, Arpan Pal
Publication date
2021/11/25
Patent office
US
Application number
17122041
Description
This disclosure relates generally to action recognition and more particularly to system and method for real-time radar-based action recognition. The classical machine learning techniques used for learning and inferring human actions from radar images are compute intensive, and require volumes of training data, making them unsuitable for deployment on network edge. The disclosed system utilizes neuromorphic computing and Spiking Neural Networks (SNN) to learn human actions from radar data captured by radar sensor (s). In an embodiment, the disclosed system includes a SNN model having a data pre-processing layer, Convolutional SNN layers and a Classifier layer. The preprocessing layer receives radar data including doppler frequencies reflected from the target and determines a binarized matrix. The CSNN layers extracts features (spatial and temporal) associated with the target's actions based on the…

My opinion only DYOR
FF

AKIDA BALLISTA

 

[Rise from the ashes]

The following link takes you to Dr. Arpan Pal’s many, many Google citations:

Dr. Arpan Pal

Distinguished Chief Scientist and Research Area Head, Embedded Systems, TCS Research, TCS - Cited by 6,390 - Embedded Systems - Computational Sensing - Signal/Image Processing and ML - Edge Computing and Edge AI

scholar.google.com

This is but one of dozens:

System and method for real-time radar-based action recognition using spiking neural network (snn)
Inventors
DEY Sounak, Arijit Mukherjee, Dighanchal Banerjee, RANI Smriti, Arun George, Tapas Chakravarty, Arijit Chowdhury, Arpan Pal
Publication date
2021/11/25
Patent office
US
Application number
17122041
Description
This disclosure relates generally to action recognition and more particularly to system and method for real-time radar-based action recognition. The classical machine learning techniques used for learning and inferring human actions from radar images are compute intensive, and require volumes of training data, making them unsuitable for deployment on network edge. The disclosed system utilizes neuromorphic computing and Spiking Neural Networks (SNN) to learn human actions from radar data captured by radar sensor (s). In an embodiment, the disclosed system includes a SNN model having a data pre-processing layer, Convolutional SNN layers and a Classifier layer. The preprocessing layer receives radar data including doppler frequencies reflected from the target and determines a binarized matrix. The CSNN layers extracts features (spatial and temporal) associated with the target's actions based on the…

My opinion only DYOR
FF

AKIDA BALLISTA

Thanks for that bookmarked for further investigation.

 

[Rise from the ashes]

Regarding my post with deep fakes the other day. I could not tell this was not Tom cruise

 

[Fact Finder]

The following link takes you to Dr. Arpan Pal’s many, many Google citations:

Dr. Arpan Pal

Distinguished Chief Scientist and Research Area Head, Embedded Systems, TCS Research, TCS - Cited by 6,390 - Embedded Systems - Computational Sensing - Signal/Image Processing and ML - Edge Computing and Edge AI

scholar.google.com

This is but one of dozens:

System and method for real-time radar-based action recognition using spiking neural network (snn)
Inventors
DEY Sounak, Arijit Mukherjee, Dighanchal Banerjee, RANI Smriti, Arun George, Tapas Chakravarty, Arijit Chowdhury, Arpan Pal
Publication date
2021/11/25
Patent office
US
Application number
17122041
Description
This disclosure relates generally to action recognition and more particularly to system and method for real-time radar-based action recognition. The classical machine learning techniques used for learning and inferring human actions from radar images are compute intensive, and require volumes of training data, making them unsuitable for deployment on network edge. The disclosed system utilizes neuromorphic computing and Spiking Neural Networks (SNN) to learn human actions from radar data captured by radar sensor (s). In an embodiment, the disclosed system includes a SNN model having a data pre-processing layer, Convolutional SNN layers and a Classifier layer. The preprocessing layer receives radar data including doppler frequencies reflected from the target and determines a binarized matrix. The CSNN layers extracts features (spatial and temporal) associated with the target's actions based on the…

My opinion only DYOR
FF

AKIDA BALLISTA

Here is another one where the processor is left open but the idea of it being personalised to the patient speaks to on chip training:

https://scholar.google.com/scholar?oi=bibs&cluster=1843484726856234880&btnI=1&hl=en

My opinion only DYOR
FF

AKIDA BALLISTA