An introduction to Hierarchical Temporal Memory and its foreseen application in ROBOFOOT and SEEGRIP

Hierarchical Temporal Memory (HTM) is a classification and pattern recognition framework modeled on the neocortex. One of the features that makes this framework different from others is its use of "time as a supervisor" to create invariant representations. For this reason, HTM models are trained with sequences of patterns (i.e. temporally correlated patterns). The underlying learning algorithms used in HTMs are not specific to particular sensory domains and thus can be applied to a broad set of problems that involve modeling complex sensory data. We will discuss possible applications of HTM in the ROBOFOOT and SEEGRIP projects.