A Physics-Inspired Optimizer: Velocity Regularized Adam

TL;DR

VRAdam is a physics-inspired optimizer that stabilizes training by dynamically damping weight updates, leading to improved performance over standard optimizers like AdamW across diverse deep learning tasks.

Contribution

Introduces VRAdam, a novel optimizer combining velocity-based regularization with Adam, supported by theoretical analysis and extensive empirical benchmarking.

Findings

VRAdam reduces oscillations and accelerates convergence.

VRAdam outperforms AdamW on multiple tasks.

Theoretical convergence rate of O(ln(N)/√N) established.

Abstract

We introduce Velocity-Regularized Adam (VRAdam), a physics-inspired optimizer for training deep neural networks that draws on ideas from quartic terms for kinetic energy with its stabilizing effects on various system dynamics. Previous algorithms, including the ubiquitous Adam, operate at the so-called adaptive edge of stability regime during training, leading to rapid oscillations and slowed convergence of loss. However, VRAdam adds a higher order penalty on the learning rate based on the velocity such that the algorithm automatically slows down whenever weight updates become large. In practice, we observe that the effective dynamic learning rate shrinks in high-velocity regimes, and damping oscillations. By combining this velocity-based regularizer for global damping with per-parameter scaling of Adam, we create a powerful hybrid optimizer. For this optimizer, we provide rigorous theoretical analysis of operation at the edge of stability from a physical and control perspective for the momentum. Furthermore, we derive convergence bounds with the rate $\mathcal{O}(\ln(N)/\sqrt{N})$ for a stochastic non convex objective under mild assumptions. We demonstrate that VRAdam exceeds the performance against standard optimizers including AdamW. We benchmark various tasks such as image classification, language modeling, and generative modeling using diverse architectures and training methodologies including Convolutional Neural Networks (CNNs), Transformers, and GFlowNets.