Reductions Paper

In “Machine Learning Techniques—Reductions Between Prediction Quality Metrics” Beygelzimer, Langford, and Zadrozny (2009?) summarize a bunch of “techniques, called reductions, for converting a problem of minimizing one loss function into a problem of minimizing another, simpler loss function.” They give a simplified overview of machine learning algorithms and sampling methods relating them to error correcting codes and regret minimization.