Bayesian Clustering of Sensory Inputs by Dynamics
This paper introduces a Bayesian method for unsupervised clustering of dynamic processes and applies it to the abstraction of sensory inputs of a mobile robot. The method starts by transforming the sensory inputs into Markov chains and then applies an agglomerative clustering procedure to discover the most probable set of clusters capturing the robot's experiences. To increase efficiency, the method uses an entropy-based heuristic search strategy.
1. Department of† Statistics, The Open University.
2. Knowledge Media Institute, The Open University.
3. Department of Computer Science, University of Massachusetts at Amherst.