-------------
Version 1.2.1
-------------

-Added gelnet.cv() to perform k-fold cross-validation for automatic parameter selection
-Simplified user inferface, especially for constructing models with a predefined number of non-zero weights
-Added a balanced loss computation to logistic regression models to use with heavily-imbalanced datasets
-Added an option for non-negativity constraint

-----------
Version 1.2
-----------

-Changed the names of L1- and L2-norm penalty coefficients to l1 and l2 in objective functions, to be consistent with their optimizers
-Added one-class regression and its kernel version: gelnet.oneclass() and gelnet.kor()
-Added an ability to fix the bias term to zero to gelnet.krr()
-Added silencing option to gelnet.klr()
-adj2nlapl() is now more stable for networks with "free-floating" nodes

-----------
Version 1.1
-----------

-Added a translation component to ridge regression
-Added kernel ridge regression: gelnet.krr()
-Added kernel logistic regression: gelnet.klr()
