Timezone: »

Learning where to learn: Gradient sparsity in meta and continual learning
Johannes von Oswald · Dominic Zhao · Seijin Kobayashi · Simon Schug · Massimo Caccia · Nicolas Zucchet · João Sacramento

Tue Dec 07 04:30 PM -- 06:00 PM (PST) @ None #None

Finding neural network weights that generalize well from small datasets is difficult. A promising approach is to learn a weight initialization such that a small number of weight changes results in low generalization error. We show that this form of meta-learning can be improved by letting the learning algorithm decide which weights to change, i.e., by learning where to learn. We find that patterned sparsity emerges from this process, with the pattern of sparsity varying on a problem-by-problem basis. This selective sparsity results in better generalization and less interference in a range of few-shot and continual learning problems. Moreover, we find that sparse learning also emerges in a more expressive model where learning rates are meta-learned. Our results shed light on an ongoing debate on whether meta-learning can discover adaptable features and suggest that learning by sparse gradient descent is a powerful inductive bias for meta-learning systems.

Author Information

Johannes von Oswald (ETH Zurich)
Dominic Zhao (ETH Zurich)
Seijin Kobayashi (ETHZ)
Simon Schug (ETH Zürich)
Massimo Caccia (MILA)
Nicolas Zucchet (ETH Zürich)
João Sacramento (ETH Zurich)

More from the Same Authors