Autonomous learning from one or several examples remains a key challenge in Machine Learning (ML). Despite recent progress in related fields such as vision and language, deep learning methodologies do not provide a satisfactory solution for learning new concepts rapidly from a few data examples.
Traditional Machine Learning techniques require large datasets to learn, often through time consuming, computationally expensive and extensive iterative training.
Ideally, Machine Learning needs to be achieved 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 technology that overcomes these problems and provides a step change in machine learning capabilities. It will significantly enhance system behaviour, reduce the computational burden and reduce the time to market for new product development or learning new features. It can handle both spatial and temporal information and can be used with our patented cognitive processing architecture. In short, it is a key enabler for a wide range of next generation Artificial Intelligent, Robotic and Autonomous Systems, systems which will underpin many global economic activities.
- Performs real time autonomous and reinforcement learning.
- Does not require large datasets or computationally intensive training.
- Mitigates the need weights and associated multipliers.
- Employs fuzzy variables, learns fuzzy rules and handles unstructured data. Hence, provides good fault tolerance and pattern recognition in noisy environments.
- Implements Fuzzy Multi-Criteria Decision Making (F-MCDM) technology.