On Enforcing Better Conditioned Meta-Learning for Rapid Few-Shot Adaptation

Inspired by the concept of preconditioning, we propose a novel method to increase adaptation speed for gradient-based meta-learning methods without incurring extra parameters. We demonstrate that recasting the optimisation problem to a non-linear least-squares formulation provides a principled way to actively enforce a wellconditioned parameter space for meta-learning models based on the concepts of the condition number and local curvature. Our comprehensive evaluations show that the proposed method significantly outperforms its unconstrained counterpart especially during initial adaptation steps, while achieving comparable or better overall results on several few-shot classification tasks -creating the possibility of dynamically choosing the number of adaptation steps at inference time. 4. We conduct in-depth analyses regarding different network architectures, ability to continue adaptation beyond the training horizon, performance regarding an extended set of N -way K-shot few-shot classification scenarios, and demonstrate the efficacy of our method on all five popular few-shot classification benchmarks.