Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Leveraging Domain Knowledge for Reinforcement Learning using MMC Architectures

: Ramamurthy, Rajkumar; Bauckhage, Christian; Sifa, Rafet; Schücker, Jannis; Wrobel, Stefan


Tetko, I.V.:
Artificial Neural Networks and Machine Learning - ICANN 2019. Deep Learning : 28th International Conference on Artificial Neural Networks, Munich, Germany, September 17-19, 2019, Proceedings, Part II
Cham: Springer International Publishing, 2019 (Lecture Notes in Computer Science 11728)
ISBN: 978-3-030-30483-6 (Print)
ISBN: 978-3-030-30484-3 (Online)
International Conference on Artificial Neural Networks (ICANN) <28, 2019, Munich>
Conference Paper
Fraunhofer IAIS ()

Despite the success of reinforcement learning methods in various simulated robotic applications, end-to-end training suffers from extensive training times due to high sample complexity and does not scale well to realistic systems. In this work, we speed up reinforcement learning by incorporating domain knowledge into policy learning. We revisit an architecture based on the mean of multiple computations (MMC) principle known from computational biology and adapt it to solve a reacher task. We approximate the policy using a simple MMC network, experimentally compare this idea to end-to-end deep learning architectures, and show that our approach reduces the number of interactions required to approximate a suitable policy by a factor of ten.