Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

Auto Neural Node and Optimization Stastistics

Accepted Solution Solved
Reply
Frequent Contributor
Posts: 75
Accepted Solution

Auto Neural Node and Optimization Stastistics

Hello,

As i saw from the results of the optimization process in the Auto Neural Node the default statistic that drives the optimization algorithm is validation misclassification. How can we select another statistic such as the validation average squared error? Is there a relevant option in the properties? I cannot find something.

Thanks in advance,

Andreas


Accepted Solutions
Solution
‎03-21-2013 05:27 PM
Occasional Contributor
Posts: 17

Re: Auto Neural Node and Optimization Stastistics

Posted in reply to andreas_zaras

Since there is an almost unlimited number of possible network configurations, EM offeres a couple of approaches to cover a wide range of needs. The neural node allows you to control a single hidden layer network.  The autoneural node provides an algorithm to build a multilayer network.  The default autoneural action is to simply train a single network to give you a baseline model. Full controll only comes from PROC NEURAL.

G

View solution in original post


All Replies
Super User
Posts: 19,839

Re: Auto Neural Node and Optimization Stastistics

Posted in reply to andreas_zaras

Under TRAIN in the left hand panel there is an option called MODEL SELECTION CRITERION. You can change that.

Frequent Contributor
Posts: 75

Re: Auto Neural Node and Optimization Stastistics

Hello,

Thanks for your answer. What you say is valid for the Neural Node. I am asking about the Auto Neural Node, where you can "play" with more elements of the ANN's architecture.

ANy ideas?

Thnaks,

Andreas

Solution
‎03-21-2013 05:27 PM
Occasional Contributor
Posts: 17

Re: Auto Neural Node and Optimization Stastistics

Posted in reply to andreas_zaras

Since there is an almost unlimited number of possible network configurations, EM offeres a couple of approaches to cover a wide range of needs. The neural node allows you to control a single hidden layer network.  The autoneural node provides an algorithm to build a multilayer network.  The default autoneural action is to simply train a single network to give you a baseline model. Full controll only comes from PROC NEURAL.

G

Frequent Contributor
Posts: 75

Re: Auto Neural Node and Optimization Stastistics

Posted in reply to adjgiulio

Thank you very much!

Frequent Contributor
Posts: 75

Re: Auto Neural Node and Optimization Stastistics

Posted in reply to adjgiulio

I am trying to mark lots of questions as answered but in some of them there is not such option. Do you know how to find such an option (mark as answered)?

Thnaks

🔒 This topic is solved and locked.

Need further help from the community? Please ask a new question.

Discussion stats
  • 5 replies
  • 760 views
  • 0 likes
  • 3 in conversation