(1 Mark] 4. Consider Cycle-GAN using which we solve the problem of Image-to-image translation where we want to learn a mapping that translate X into Y and Y into X. Consider trying to train this network using only the GAN losses that are defined as: (CAN(F, D.) Ex [log D. (x)]+ Ex llog(1 2,(F\YDI baan( GD) Tlog Dy Ex[log(1 HD,GX) Find the optimal discriminators Do and Df that maximize (gan(E, D) and PGAN (GI Dk a respectively, Then, find the optimal generators Fland G that minimize EGAN(F, DX) and (GAN(G,D1), respectively. 1242 Marks] G(x) D.(x) 2 ~ Pdata (x) G 2D.(F(y)) F(y) D, (G(x)) Y ~ Pdata (y) F Dr Dyly) (1 Mark] 4. Consider Cycle-GAN using which we solve the problem of Image-to-image translation where we want to learn a mapping that translate X into Y and Y into X. Consider trying to train this network using only the GAN losses that are defined as: (CAN(F, D.) Ex [log D. (x)]+ Ex llog(1 2,(F\YDI baan( GD) Tlog Dy Ex[log(1 HD,GX) Find the optimal discriminators Do and Df that maximize (gan(E, D) and PGAN (GI Dk a respectively, Then, find the optimal generators Fland G that minimize EGAN(F, DX) and (GAN(G,D1), respectively. 1242 Marks] G(x) D.(x) 2 ~ Pdata (x) G 2D.(F(y)) F(y) D, (G(x)) Y ~ Pdata (y) F Dr Dyly)