Adversarial Residual Transform Network (ARTN) の実装に関するメモ（２） - Qiita

# generate toy data n_s = 1000 n_t = 1000 n_classes = 3 p_s = [0.4, 0.3, 0.3] p_t = [0.4, 0.3, 0.3] mu_s_1 = [0.0, 0.0] sig_s_1 = [0.2, 0.2] mu_s_2 = [0.0, 1.0] sig_s_2 = sig_s_1 mu_s_3 = [0.5, 0.5] sig_s_3 = sig_s_1 # pattern 1 # mu_t_1 = [2.0, 0.5] # sig_t_1 = sig_s_1 # mu_t_2 = [2.0, 1.5] # sig_t_2 = sig_s_1 # mu_t_3 = [2.5, 1.0] # sig_t_3 = sig_s_1 # pattern 2 # mu_t_1 = [0.5, 0.5] # sig_t_1 =