Sharpness-Aware Minimization for Efficiently Improving Generalization
This addresses the generalization issue in machine learning for practitioners using overparameterized models, offering a novel optimization approach that is not incremental.
The paper tackles the problem of poor generalization in overparameterized models by introducing Sharpness-Aware Minimization (SAM), a procedure that minimizes both loss value and loss sharpness, leading to improved generalization across multiple benchmark datasets and models, with state-of-the-art performance in some cases and robustness to label noise.
In today's heavily overparameterized models, the value of the training loss provides few guarantees on model generalization ability. Indeed, optimizing only the training loss value, as is commonly done, can easily lead to suboptimal model quality. Motivated by prior work connecting the geometry of the loss landscape and generalization, we introduce a novel, effective procedure for instead simultaneously minimizing loss value and loss sharpness. In particular, our procedure, Sharpness-Aware Minimization (SAM), seeks parameters that lie in neighborhoods having uniformly low loss; this formulation results in a min-max optimization problem on which gradient descent can be performed efficiently. We present empirical results showing that SAM improves model generalization across a variety of benchmark datasets (e.g., CIFAR-10, CIFAR-100, ImageNet, finetuning tasks) and models, yielding novel state-of-the-art performance for several. Additionally, we find that SAM natively provides robustness to label noise on par with that provided by state-of-the-art procedures that specifically target learning with noisy labels. We open source our code at \url{https://github.com/google-research/sam}.