Output details

Improving reservoirs using intrinsic plasticity

<24> The paper proposes a biologically motivated learning mechanism to improve reservoir neural networks, which is based on the paradigm of intrinsic plasticity that has been introduced by Steil to the recurrent domain in 2004. This work in reservoir computing was key to a EU-FP7 large-scale project AMARSi that is coordinated by Steil (10 partners, 6 countries, 7 million Euro funding, 4 years duration, which after 3 years produce more than 140 publications). AMARSi also includes co-author Schrauwen and applies reservoir computing to movement learning in the robotics domain.