Multi-Label Prediction via Compressed Sensing

Daniel Hsu, Sham M. Kakade, John Langford +1

We consider multi-label prediction problems with large output spaces under the assumption of output sparsity -- that the target (label) vectors have small support. We develop a gen…