The folder containing this file contains the code used for running MK-MTSFL and MK-MTSFL of Multi-task Multiple Kernel Learning

The regression folder contains the regression codes and the classification folder contains the classification code. 
Libsvm 2.9.1 matlab interface (http://www.csie.ntu.edu.tw/~cjlin/libsvm/) has been used in classification for both MK-MTFL and MK-MTSFL.
CVX 1.21 (http://cvxr.com/cvx/)  is needed for running the code of MK-MTSFL.

Some comments:

1. A wrapper will be needed to call these codes. The wrapper should also send the adequate inputs in the format prescribed in each file.

For MK-MTFL

regression
trrrg_mtmkl.m :  Code for training the model
terrg_mtmkl.m : Code for prediction

classification
trcl_mtmkl.m : Code for training the model
tecl_mtmkl.m : Code for prediction

For MK-MTSFL
solveGrVarProb.m : Code calls CVX to solve the intermediate formulation

regression
trrg_mkmtfl.m : Code for training the model
terg_mkmtfl.m : Code for prediction

classification
trcl_mkmtfl.m : Code for training the model
tecl_mkmtfl.m : Code for prediction

For any queries, feel free to mail at

saketh@cse.iitb.ac.in
pratik.j@cse.iitb.ac.in

