Transferring Knowledge across Learning Processes

Abstract

In complex transfer learning scenarios new tasks might not be tightly linked to previous tasks. Approaches that transfer information contained only in the final parameters of a source model will therefore struggle. Instead, transfer learning at a higher level of abstraction is needed. We propose Leap, a framework that achieves this by transferring knowledge across learning processes. We associate each task with a manifold on which the training process travels from initialization to final parameters and construct a meta-learning objective that minimizes the expected length of this path. Our framework leverages only information obtained during training and can be computed on the fly at negligible cost. We demonstrate that our framework outperforms competing methods, both in meta-learning and transfer learning, on a set of computer vision tasks. Finally, we demonstrate that Leap can transfer knowledge across learning processes in demanding reinforcement learning environments (Atari) that involve millions of gradient steps.

arXiv commentPublished as a conference paper at ICLR 2019; 23 pages, 8 figures, 6 tables

Cited by in corpus (12)

• Knowledge Distillation: A Survey
• Empirical Bayes Transductive Meta-Learning with Synthetic Gradients
• Meta-Learning with Warped Gradient Descent
• Meta-Learning across Meta-Tasks for Few-Shot Learning
• Meta-Learning Acquisition Functions for Transfer Learning in Bayesian Optimization
• Modular Meta-Learning with Shrinkage
• Principal component analysis for Gaussian process posteriors
• Transfer Learning with Gaussian Processes for Bayesian Optimization
• Graceful Degradation and Related Fields
• Multilingual Speech Recognition using Knowledge Transfer across Learning Processes
• Large-Scale Meta-Learning with Continual Trajectory Shifting
• Bootstrapped Meta-Learning