ISL talk about a patent for an AI acceptance methodology that allows for DoD testing of AI solutions using essentially the same statistically based methodology in use today.
https://www.islinc.com/national-security/artificial-intelligence
Artificial Intelligence, Neuromorphic Computing and DoD Acceptance Testing
I
SL 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.
Additionally, ISL has pioneered an AI acceptance methodology that allows for DoD testing of AI solutions using essentially the same statistically based methodology in use today. ISL was awarded a US Patent for this February (see link). The methodology leverages ISL’s RF Digital Engineering tools (https://www.islinc.com/digital-engineering ).
US11256988B1 Process and method for real-time sensor neuromorphic processing
Priority date: 20210719
View attachment 19555
A
novel system and method are described that allows for implementation of compact and efficient deep learning AI solutions to advanced sensor signal processing functions. The process includes the following stages: (1) A method for generating requisite annotated training data in sufficient quantity to ensure convergence of a deep learning neural network (DNN); (2) A method for implementing the resulting DNN onto a Spiking Neural Network (SNN) architecture amenable to efficient neuromorphic integrated circuit (IC) architectures; (3) A method for implementing the solution onto a neuromorphic IC; and (4) A statistical method for ensuring reliable performance.
[003] ... Neural networks work with functionalities similar to the human brain. The invention includes both a training cycle and a live (online) operation. The training cycle includes five elements and comprises the build portion of the deep learning process. The training cycle requirements ensure adequate convergence and performance. The live (online) operation includes the live operation of
a Spiking Neural Network (SNN) designed by the five steps of the training cycle. The invention is part of a new generation of neuromorphic computing architectures, including Integrated Circuits (IC).
This new generation of neuromorphic computing architectures includes IC, deep learning and machine learning.
[008] ... A
high-fidelity (hi-fi) sensor model can be used to generate the requisite training data and/or training environment. The sensor model (in this case RFView®) (https://RFView.ISLinc.com is used to generate training data in sufficient quantity to ensure convergence of the DNN neuron weights. Thereafter, a suitable DNN interface is established wherein the raw sensor training cycle data is preprocessed into a format suitable for presentation to a DNN. In this step of the method, the DNN is converted to an SNN. Discussion regarding the conversion from DNN to SNN is found in the paper by M. Davies et al., “Advancing neuromorphic Computing with Loihi: A survey of results and outlook,” Proceedings of the IEEE , vol. 109, no. 5, pp. 911-934, 2021. The SSN is then implemented on a suitable neuromorphic architecture or IC to achieve requisite performance.
[sotto voce: "I'm not at all sure Ella would be in raptures about the specification"]
#########################################################################################
It seems likely that this patent was conceived before ISL and BrainChip got together because of the long lead time to develop the system, and because of the complex process of developing the weight data in DNN and converting to SNN:
Information Systems Labs Joins BrainChip Early Access Program
Laguna Hills, Calif. – January 9, 2022
The system requires the development of the training data in DNN format "
in sufficient quantity to ensure convergence of the DNN neuron weights", and the conversion of the training data to SNN. This would be unnecessary in a system designed from the ground up for Akida.
ISL submitted their idea in July 2022, so there would have been time to incorporate Akida into their proposal.
for them and you 
