Implement some ML algorithms in scala.

1. base bgd, sgd, mbgd to get started.
2. Frequently-used GD algos：
   • SGD with momentum and NAG
   • AdaGrad
   • AdaDelta
   • RMSProp
   • Adam with AMSGrad
   • AdaMax
   • Nadam
   • FTRL_Proximal(online learning)
   • SWATS
3. Quasi-Newton Methods
   • DFP with golden section method
   • BFGS with golden section method
   • CG with Fletcher-Reeves method
   • L-BFGS (implement with line search method of Wolfe-Powell)
4. common loss function for binary classification.
5. algos in package newml:
   • redesign ml for multi-classification

note:

1. the package ml is just tested for binary classification(especially logloss) now.