19:36 < luffy1996> https://www.irccloud.com/pastebin/bbxGYtBA/

19:37 < luffy1996> @zoq: @rcurtin: I am trying to use categorical crossentropy on softmax layer in a ann network. I am unsure about how to integrate softmax layer in ann. I see that LogSoftMax is available with mlpack under ann. SoftmaxRegression is also available in the methods. Would you please help me out how to proceed for softmax in the last layer of ann. Something like

19:37 < luffy1996>

19:37 < luffy1996> model.Add<SoftMax<> >();

19:37 < luffy1996> What I want is a vector showing probability of a specific class. Then using categorical crossentropy to compute gradients.

19:43 < zoq> luffy1996: The easiest solution is probably to implement a softmax layer and pass the layer, see https://github.com/mlpack/mlpack/blob/4bd01bbc98889e1ade49302b79d791275854be37/src/mlpack/tests/feedforward_network_test.cpp#L575 for how to use a custom layer.

19:47 < luffy1996> Thanks I will look into it tomorrow. Its late night here so going to get some sleep. Thanks :)

19:48 < zoq> Sounds good, sleep well.

23:37 < manthan> luffy1996 : let me know if you need some help, i have experience with implementing layers for ann.