Single Flexible Architecture – Multiple Applications
Our aim is to develop Machine Learning technology that is genuinely useful, that solves real problems and makes a positive economic and social impact. We think the best way to achieve this is to create intelligent products that can learn and improve over time.
Cyceera’s Machine Learning IP is modular, scalable and can be combined with our cognitive processing technology allowing it to target a wide range of AI applications of varying complexity.
To fulfill Machine Learning’s true potential a much more efficient Machine Learning algorithm/implementation is required. Ideally, Machine Learning needs to be achieved autonomously using only one (one-shot) or a few examples and in real time. Cyceera has invented an efficient real time (non neural network) Machine Learning/inference technology, which can enhance system behaviour, reduce the computational burden and reduce the time to market for new product development or learning new features. Cyceera’s Autonomous Real Time Machine Learning (ARTML) technology has a high functional density, a small silicon footprint and consumes little power.
As the common denominator in all these systems, Cyceera’s scalable Machine Learning technology provides an attractive solution that will significantly enhance future product development and performance. Low power consumption and a small silicon footprint allow our Machine Learning technology to be integrated into all manner of gadgets, wearables and sensors enabling them to operate intelligently and independently. Such intelligent sensors will drive the internet of things (IoT) forward.
Benefits
- Real time autonomous Learning – automatically extracts features
- Simple digital reconfigurable architecture that target different applications
- Modular, scalable, power efficient– small silicon footprint
- Explainable AI
Traits of Conventional Neural Networks
- Application specific
- Require large datasets to learn, often through time consuming, computationally expensive and extensive iterative training.
- Rely on cloud services
- Supervised learning requiring labelled datasets
- Run on large, power hungry CPUs, GPUs and Field Programmable Gate Arrays (FPGAs) and are therefore not applicable to small low power portable applications
- Black boxes that cannot explain derivation of results