Making sense of the sensory data - coordinate systems by hierarchical decomposition
Having the right sensory channels is an important ingredient for building an autonomous agent, but we still have the problem of making sense of the sensory data for the agent. This is the basic problem of artificial intelligence. Here we propose an algebraic method for generating abstract coordinate system representations of the environment based on the agent’s actions. These internal representations can be refined and regenerated during the lifespan of the agent.