Skip directly to content

Experiments with Reservoir Computing on the road sign problem

on Thu, 09/29/2016 - 19:51
TitleExperiments with Reservoir Computing on the road sign problem
Publication TypeConference Proceedings
Year of Conference2007
AuthorsAntonelo EA, Schrauwen B, Stroobandt D
Conference NameProceedings of the VIII Brazilian Congress on Neural Networks (CBRN)

The road sign problem is tackled in this work with Reservoir Computing (RC) networks. These networks are made of a fixed recurrent neural network where only a readout layer is trained. In the road sign problem, an agent has to decide at some point in time which action to take given relevant information gathered in the past. We show that RC can handle simple and complex T-maze tasks (which are a subdomain of the road sign problem).