In the model architecture, the "MLPVanilla" class inserts 2 hidden, linear layers. It seems like you don't have a "Softmax" layer at the end to obtain the survival probability distribution. Is this because it is already included in the loss function or do we not need to use it at all ?
Hi,
I am following this DeepHit tutorial for a single event - https://github.com/havakv/pycox/blob/master/examples/deephit.ipynb
In the model architecture, the "MLPVanilla" class inserts 2 hidden, linear layers. It seems like you don't have a "Softmax" layer at the end to obtain the survival probability distribution. Is this because it is already included in the loss function or do we not need to use it at all ?
net = tt.practical.MLPVanilla(in_features, num_nodes, out_features, batch_norm, dropout)
model = DeepHitSingle(net, tt.optim.Adam, alpha=0.2, sigma=0.1, duration_index=labtrans.cuts)
Also, it seems like you are not implementing "residual connections " as mentioned in the DeepHit paper. Could you please explain the reason for this ?
Thanks Ani