The Advantage of Conditional Meta-Learning for Biased Regularization and Fine Tuning

Neural Information Processing Systems 

Biased regularization and fine tuning are two recent meta-learning approaches.