Continual Meta-Learning

Meta-Gradient Optimization

Technique that uses gradients of gradients to dynamically adjust model optimization parameters during continual learning, thereby improving its adaptation capacity.

← Terug