Using Statistics to Automate Stochastic Optimization
This addresses the bottleneck of manual learning rate tuning for machine learning practitioners, offering an incremental improvement over existing adaptive optimizers.
The paper tackles the problem of tuning learning rates in stochastic gradient methods by proposing a statistical test to detect when progress stops, then dropping the learning rate automatically. Experiments on deep learning tasks show that this method matches the performance of hand-tuned approaches with default parameters.
Despite the development of numerous adaptive optimizers, tuning the learning rate of stochastic gradient methods remains a major roadblock to obtaining good practical performance in machine learning. Rather than changing the learning rate at each iteration, we propose an approach that automates the most common hand-tuning heuristic: use a constant learning rate until "progress stops," then drop. We design an explicit statistical test that determines when the dynamics of stochastic gradient descent reach a stationary distribution. This test can be performed easily during training, and when it fires, we decrease the learning rate by a constant multiplicative factor. Our experiments on several deep learning tasks demonstrate that this statistical adaptive stochastic approximation (SASA) method can automatically find good learning rate schedules and match the performance of hand-tuned methods using default settings of its parameters. The statistical testing helps to control the variance of this procedure and improves its robustness.