+ 2 points Someone is trying to learn the data above using logistic regression. It is clearly linearly separable data. So we can assume that training error achieved will be zero if considered only cross-entropy loss function. However, here is some person using the following loss function. What do you think of the training error (increases/decreases/remains same) for the following situations: CO = 0, C1 = 10000, C2=0. m L(x,y, W) = -(y(i)log(90))+ (1 - y(i))log(1 - y(i))) + Cow +C w + C2w i=1 O Remains Same O Increases Decreases 2 points Someone is trying to learn the data above using logistic regression. It is clearly linearly separable data. So we can assume that training error achieved will be zero if considered only cross-entropy loss function. However, here is some person using the following loss function. What do you think of the training error (increases/decreases/remains same) for the following situations: CO = 0, C1 = 0, C2=10000. L(2, 4,W) = -Zy(i)log(i(i)) + (1 y(i))log(1 (i))) + Cow +C\w{+Cyw i=1 O Remains Same O Increases Decreases