Identification and removal of contaminants in sEMG recordings through a methodology based on Fuzzy Inference and Actor-Critic Reinforcement learning Documento