Cookie Monster

Next generation Machine Learning is goal-based and adapts in real time.

This demo gives you a taste of what React AI’s Enki engine can do

The Enki engine controls the cookie monster using closed-loop analytics to guide it through its world by taking actions (back/forward, left/right) to maximise its rewards (cookies). When it was first started the engine had no information about its environment. It did not know the configuration, the rules, or the goal of the game, it had to learn it all.  More importantly, we didn’t have to tell it about the messages it would receive, nor configure the fields. We simply told it it to pick actions to maximise the score, and let it go from there.

The Cookiemonster Game

What's under the React AI hood?

The BRAIN2 engine is leveraging an innovative combination of several machine learning techniques (feed forward neural nets, recurrent neural nets, reinforcement learning and data fusion) to extract patterns from the message streams and build a holistic predictive model of the cookie monster’s world.

Data fusion is the integration of multiple varied sources of information to reduce noise and improve understanding. BRAIN2 was designed to merge messages from different sources on different timescales, for example the cookie monster’s sense of touch, movement, and sight. Touch events happen irregularly, movement messages depend on the input of the BRAIN2 engine or of the player, while the cookie monster has continuous sight of its environment.

Closed Loop Analytics is a Machine Learning technique that relies on a live feedback loop between a Machine Learning engine and the real world. The BRAIN2 engine suggests what it believes to be the best action for the monster to be happy (obtain rewards).  BRAIN2 then uses the feedback from testing these suggestions to learn and correct its model of the world.