Adding reinforcement learning features to the neural-gas method

M. Winter, G. Metta, G. Sandini
2000 Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium  
We propose a new neural approach for approximating function using a reinforcement-type learning: each time the network generates an output, the environment responds with the scalar distance between the delivered output and the expected one. Thus, this distance is the only information the network can use to modify the estimation of the multi-dimensional output. This reinforcement feature is embedded in a neuralgas method, taking advantages of the different facilities it offers. We detail the
more » ... al algorithm and we present some simulation results in order to show the behaviour of the developed method. R the approximation of the function X R at iteration k, we defined:
doi:10.1109/ijcnn.2000.860827 dblp:conf/ijcnn/WinterMS00a fatcat:5hpq7yhpprf4lf64gjykbiamge