The motivation of this project is based on the idea to apply the unsupervised learning algorithm Slow Feature Analysis (SFA) to interactive scenarios. This idea is based on the experience that SFA was successfully used in various (passive) analysis tasks that closely relate to interactive control scenarios, e.g. learning place cells, identifying objects invariant under spacial transformations, blind source separation, visual tasks like face recognition.
We recognize predictability as a crucial feature for tackling interactive scenarios, as these require estimation of consequences of possible actions.
Every organism in an environment, whether biological, robotic or virtual, must be able to predict certain aspects of its environment in order to survive or perform whatever task is intended. It needs a model that is capable of estimating the consequences of possible actions, so that planning, control, and decision-making become feasible. Predictable features are highly relevant for modeling, because predictability is a desired property of the needed consequence-estimating model by definition.
This motivates Predictable Feature Analysis (PFA), an algorithm strongly inspired by SFA: while SFA selects features by slowness, PFA selects them by predictability.
Singular Sturm-Liouville Problems with Zero Potential (q=0) and Singular Slow Feature Analysis