Landry Steve Noulawe Tchamanbe ; Paulin MELATAGIA YONTA - Algorithms to get out of Boring Area Trap in Reinforcement Learning

arima:6748 - Revue Africaine de la Recherche en Informatique et Mathématiques Appliquées, July 2, 2021, Volume 34 - 2020 - Special Issue CARI 2020 - https://doi.org/10.46298/arima.6748
Algorithms to get out of Boring Area Trap in Reinforcement Learning

Authors: Landry Steve Noulawe Tchamanbe ; Paulin MELATAGIA YONTA

Reinforcement learning algorithms have succeeded over the years in achieving impressive results in a variety of fields. However, these algorithms suffer from certain weaknesses highlighted by Refael Vivanti and al. that may explain the regression of even well-trained agents in certain environments : the difference in variance on rewards between areas of the environment. This difference in variance leads to two problems : Boring Area Trap and Manipulative consultant. We note that the Adaptive Symmetric Reward Noising (ASRN) algorithm proposed by Refael Vivanti and al. has limitations for environments with the following characteristics : long game times and multiple boring area environments. To overcome these problems, we propose three algorithms derived from the ASRN algorithm called Rebooted Adaptive Symmetric Reward Noising (RASRN) : Continuous ε decay RASRN, Full RASRN and Stepwise α decay RASRN. Thanks to two series of experiments carried out on the k-armed bandit problem, we show that our algorithms can better correct the Boring Area Trap problem.


Volume: Volume 34 - 2020 - Special Issue CARI 2020
Published on: July 2, 2021
Submitted on: September 1, 2020
Keywords: k-armed bandit,ASRN,Boring Area Trap,Reinforcement Learning,k-armed bandit,bandit à k bras.,ASRN,Piège de la Zone Ennuyeuse,Apprentissage par renforcement,bandit à k bras.,[INFO]Computer Science [cs],[INFO.INFO-AI]Computer Science [cs]/Artificial Intelligence [cs.AI],[STAT.ML]Statistics [stat]/Machine Learning [stat.ML]


Share