Hi all,

looking at the nice multi layer perceptron feature I realized we have
different APIs for defining a cost function.
In the regression package we have
http://svn.apache.org/repos/asf/hama/trunk/ml/src/main/java/org/apache/hama/ml/regression/CostFunction.java

while in perception (shouldn't it be perceptron?) we have:
http://svn.apache.org/repos/asf/hama/trunk/ml/src/main/java/org/apache/hama/ml/perception/CostFunction.java

then for a logistic regression model (which encapsulates a logistic cost
function) in regression we have:
http://svn.apache.org/repos/asf/hama/trunk/ml/src/main/java/org/apache/hama/ml/perception/LogisticCostFunction.java

while in perception we have:
http://svn.apache.org/repos/asf/hama/trunk/ml/src/main/java/org/apache/hama/ml/perception/LogisticCostFunction.java

so I think it'd be good if we can abstract and define a unified API for
cost function, models, etc. in the machine learning package.

What do you think?

Thanks,
Tommaso

Reply via email to