Scalable Nested Optimization for Deep Learning
This addresses a bottleneck in applying nested optimization to large-scale deep learning problems, though it appears incremental as it builds on existing nested optimization concepts.
The paper tackles the problem of scaling nested optimization (e.g., for hyperparameter optimization and GANs) in deep learning, where classical methods often fail at large scales, and builds tools to address this scalability issue.
Gradient-based optimization has been critical to the success of machine learning, updating a single set of parameters to minimize a single loss. A growing number of applications rely on a generalization of this, where we have a bilevel or nested optimization of which subsets of parameters update on different objectives nested inside each other. We focus on motivating examples of hyperparameter optimization and generative adversarial networks. However, naively applying classical methods often fails when we look at solving these nested problems on a large scale. In this thesis, we build tools for nested optimization that scale to deep learning setups.