Model-based control of a robot manipulator
Model-based control of a robot manipulator
Neural networks and the bias/variance dilemma
Neural Computation
Arithmetic coding for data compression
Communications of the ACM
Artificial Intelligence Review - Special issue on lazy learning
Stochastic Complexity in Statistical Inquiry Theory
Stochastic Complexity in Statistical Inquiry Theory
Incremental Online Learning in High Dimensions
Neural Computation
Hi-index | 0.00 |
Incremental learning of sensorimotor transformations in high dimensionalspaces is one of the basic prerequisites for the success of autonomous robot devices as well as biological movement systems. So far, due tosparsity of data in high dimensional spaces, learning in such settingsrequired a significant amount of prior knowledge about the learning task,usually provided by a human expert. In this paper we suggest a partialrevision of the view. Based on empirical studies, we observed that, despitebeing globally high dimensional and sparse, data distributions fromphysical movement systems are locally low dimensional and dense.Under this assumption, we derive a learning algorithm, Locally AdaptiveSubspace Regression, that exploits this property by combining a dynamicallygrowing local dimensionality reduction technique as a preprocessing stepwith a nonparametric learning technique, locally weighted regression, thatalso learns the region of validity of the regression. The usefulness of thealgorithm and the validity of its assumptions are illustrated for asynthetic data set, and for data of the inverse dynamics of human armmovements and an actual 7 degree-of-freedom anthropomorphic robot arm.