Predictive Top-down Knowledge Improves Neural Exploratory Bottom-up Clustering

In this paper we propose an approach to robot learning by imitation that uses the multimodal inputs of language instruction, vision and motor. In our approach a student robot learns from a teacher robot how to perform three separate behaviours, 'pick', 'lift' and 'go' based on these inputs. We considered two neural architectures for performing this robot learning. First, a one-step architecture trained with two different learning approaches either based on Kohonen's self-organising map or based on the Helmholtz machine turns out to be inefficient or not capable of performing differentiated behaviour. In response we produced a hierarchical architecture that combines both learning approaches to overcome these problems. In doing o the proposed robot system models specific aspects of learning using concepts of the mirror neuron system and the hierarchical organisation of the motor system with regards to demonstration learning.

@InProceedings{HWS04, 
 	 author =  {Hung, Chihli and Wermter, Stefan and Smith, Peter},  
 	 title = {Predictive Top-down Knowledge Improves Neural Exploratory Bottom-up Clustering}, 
 	 booktitle = {Proceedings of ECIR’04 European Conference on Information Retrieval},
 	 editors = {},
 	 number = {},
 	 volume = {},
 	 pages = {154--166},
 	 year = {2004},
 	 month = {Apr},
 	 publisher = {Springer},
 	 doi = {}, 
 }