chcorbi/MultiTaskLearning: Package to apply MTL on a few dataset · GitHub

#Multi-Task Learning Authors : Charles Corbière, Hamza Cherkaoui

This package implement differents multi-task learning models:

• Multilearning SVM (svm): an SVM is learning for each task
• Alternating Structure Optimization (aso): a modele assuming every task shared a low dimensional structure
• Convex Alternating Structure Optimization (caso): convex relaxation of ASO
• Clustered Multi-task Learning (cmtl): a modele assuming tasks are groupd within clusters.

Dataset included:

• a clustered toy dataset (toy)
• School data (school)
• Sarcos data (sarcos)

• To compute score for a given algorithm on a given dataset, for a test size proportion and a number of splits

python computeScores.py school cmtl 5 0.30

Here, we run 5 times CMTL on school dataset with a 30% test size proportion.

• To plot all algorithms scores for a given dataset and a number of splits, iterating on the test size proportion

python plotResults.py school 5

Here, we run 5 times for each algorithm on school dataset. Note that on current implementation, the test size range is [0.30, 0.40, 0.50, 0.60]