A Low-Cost Ethics Shaping Approach for Designing Reinforcement Learning Agents

One idea to achieve incorrect but also unethical decisions. Reinforcement learning such goal is to collect enough ethical behavior data of human (Sutton and Barto 1998) is designed to tackle intricate acting toward the given goal, and then apply the inverse real-world problems in rather short time (Strehl et al. 2006; reinforcement learning (IRL) (Amin and Singh 2016; Evans, Brafman and Tennenholtz 2002) with a performance bound Stuhlmüller, and Goodman 2016; Ng, Russell, and others (Strehl, Li, and Littman 2009); however, it relies heavily on 2000; Sezener 2015) technique to learn an ethical agent that the quality of the reward functions provided as the inputs.