The Flek Machine is a specialized Probability Machine, it represents a natural step in the AI evolution towards Real-AI.
What is a probability machine?
A Probability Machine is a special kind of ML engine that automatically builds a probabilistic modes from semi-structured data. It is an invention and a major contribution to the field of probability.
Why build a Probability Machine?
Actually, why not build a probability machine and why did it take so long?
Today there are machines that do all sorts of work. Thus, it is a natural step in AI evolution to build a learning machine that takes us beyond machine learning. Given the history of probability and its profound influence on statistics and mathematical analysis, we believe that the first step towards learning machines is to build a Probability Machine.
How does it work?
Contrary to treating models as algorithms or formulas, models built and managed by a probability machine act like database models – i.e. they can be updated, queried and explored for analytical purposes or for making prediction. Also, instead of working with probabilities as distributions, the machine treats them as first class objects – Nuggets – which can be interpreted, stored, fetched and mined. With these Nuggets, developers can work at a higher level of abstraction and develop Algorithmic Probability Programs capable of complex probabilistic computations.