Learning to Weight Imperfect Demonstrations

                  Yunke Wang 1 Chang Xu 2 Bo Du 1 Honglak Lee 3 4

              Abstract                  any access to reward signal, has achieved great success in
                                   many sequential decision making problems (Stadie et al.,
    This paper investigates how to weight imperfect        2017; Ermon et al., 2015; Finn et al., 2016). Compared
    expert demonstrations for generative adversarial        to complex reward engin ...