“V. CONCLUSIONDeveloping a real-time and power-efficient...

“V. CONCLUSION

Developing a real-time and power-efficient hardware- based implementation of a pattern-recognition engine has been of paramount importance for e-nose systems. This paper presents an application of an event-based neuromorphic approach for the encoding and classification of e-nose data. The proposed solution consists of a data-to-event encoder that implements AERO and SF-based event encoding of continuous multi-variate sensor array responses and a two-layer SNN-based classifier that implements bio-inspired learning to identify and classify repeating patterns in the event-encoded sensor responses. When applied to the MedNose dataset, the SNN-based classifier identified ten different bacteria types with a classification accuracy of 97.42%. The system outperforms the accuracy of previous implementations, which require a 20-sample window and achieve 96.2% classification accuracy [2], whilst only requiring a single sample for classification. The SNN model was validated on the Akida NSoC and recorded a dynamic power consumption of 24.5 mW and a throughput of 181 inferences per second. Such a system could significantly speed up disease diagnosis in the real world whilst achieving state-of-the-art classification performance“

https://brainchip.com/wp-content/uploads/2023/11/Bacteria-in-Blood-research-paper_Final.pdf

Say whatever you want but the so called analyst who rubbished AKIDA technology and stated “what’s a neuromorphic chip anyway” has had his epitaph already written ‘Here lies a fool of the first order.”

No other technology on the market comes within a bulls roar of being able to match the performance of even the first generation AKIDA technology.

My opinion only DYOR

Fact Finder