MTIL
Multi-Task Feature Interaction Learning (MTIL)
Publications:
- Multi-Task Feature Interaction Learning.
Kaixiang Lin, Jianpeng Xu, Inci M. Baytas, Shuiwang Ji, Jiayu Zhou
KDD 2016. [Paper]
Acknowledgement
This project is based in part upon work supported by the National Science Foundation under Grant IIS-1565596 and Office of Naval Research N00014-14-1-0631.
Folder Structure:
|>MTLconvx: convex formulations
---- MTIL_L_Lc
---- MTIL_L_S
---- MTIL_S_Lc
---- MTIL_S_S
|>MTLnonconvx: nonconvex formulations
---- MTIL_L_Ln
---- MTIL_S_Ln
|>MTLlowrankW: multi-task baseline
---- MTL_L
|>STL_Least: single-task baseline
---- RR
---- STIL
|>cfiles: mex files for accelerating computation
|>datas: synthetic sample data
|>sparsa: PNOPT solver
|>utilities: utilities files for vectorizing matrics, etc.
script_run_cell_mtl_syn.m <--- run this to get a sample results.
configurefile.m <--- tune parameters here.
PS:
It's currently the code example for all methods in the paper, we will integrate this to the multi-task learning package MALSAR soon.