Generalization in Transfer Learning

Suzan Ece Ada, Emre Ugur, H. Levent Akin

Agents trained with deep reinforcement learning algorithms are capable of performing highly complex tasks including locomotion in continuous environments. We investigate transferri…