search

CogComp / saul

Saul : Declarative Learning-Based Programming
Other
64 stars 18 forks source link

Why two jointTrains? #278

Open danyaljj opened 8 years ago

danyaljj commented 8 years ago

Why do we have objects for joint training? https://github.com/IllinoisCogComp/saul/blob/master/saul-core/src/main/scala/edu/illinois/cs/cogcomp/saul/classifier/JointTrain.scala https://github.com/IllinoisCogComp/saul/blob/master/saul-core/src/main/scala/edu/illinois/cs/cogcomp/saul/classifier/JoinTrainSparseNetwork.scala

How are they different? Do we want to somehow merge them?

kordjamshidi commented 8 years ago

I started it using the binary case (training binary classifiers jointly) and there we have the first JoinTrain. Then I wrote a more flexible version with SparseNetworks for training multi-class classifiers jointly. So their base Learner is different. But I can work on merging them. I think having both still is good just we need to merge them such that we use the type of classifiers as a parameter.