Which activation or transfer function is suitable to classify 3 type of data?

I have Iris dataset I changed the classes from flower name to 0 , 1, -1 which transfer function is suitable to use when doing manual training does logsig make the work ???

Answer 1

[Greg Heath]

Use

     [ 1 0 0
      0 1 0 
      0 0 1 ]

Hope this helps.

Thank you for formally accepting my answer

Greg

[Greg Heath]

WHOOPS!!!

My answer was given assuming you would be using a classification function like Patternnet.

I don't understand what you mean by classification by "hand"

Greg

[Samisam]

I mean I'm trying to get the output like

y = logsig(logsig(trin*iw'+repmat(b1',size(trin,1),1))*lw'+repmat(b2',size(trin,1),1));

logsig as I know gives 0 or 1 out put I want an activation function gives me 0 or 1 or -1 thanks

[Samisam]

by the way I'm using matlab R2015 32bit

Answer 2

[Greg Heath]

My computer does not allow me to use the comment box. I can't even use the full ANSWER box!

Use tansig/tansig

Thank you for formally accepting my answer

Greg

P.S. tansig == tanh !!!

[Samisam]

I tried tanh and did give the same classification for 15 sample each 5 of them belong to a different class

[Greg Heath]

I'm not sure what that means. Please elucidate.

Greg

[Samisam]

Sorry for being late I'm trying to train a MLP to classify Iris dataset using ABC. From the code of ABC I get the optimal weights and biases I entered them to the code of testing it gave me 5 right classification and 10 wrong I think the fault in my code is in training process could you give me your email to send you my code ???? thanks any way

[Salma Hassan]

dr Greg Heath can you check my question plz how does it possible in convnet, High accuracy in both of validation and test the same has 97% with error loss function 0.01

[Greg Heath]

Sorry, I cannot help.

Greg