Minibatching Offers Improved Generalization Performance for Second Order Optimizers
This work addresses the computational efficiency and tuning challenges for machine learning practitioners by showing that second-order methods with minibatches can reduce hyperparameter tuning time.
The study investigated how batch size affects the performance of second-order optimizers in training deep neural networks, finding that minibatching improves peak accuracy and reduces variance compared to full batch training.
Training deep neural networks (DNNs) used in modern machine learning is computationally expensive. Machine learning scientists, therefore, rely on stochastic first-order methods for training, coupled with significant hand-tuning, to obtain good performance. To better understand performance variability of different stochastic algorithms, including second-order methods, we conduct an empirical study that treats performance as a response variable across multiple training sessions of the same model. Using 2-factor Analysis of Variance (ANOVA) with interactions, we show that batch size used during training has a statistically significant effect on the peak accuracy of the methods, and that full batch largely performed the worst. In addition, we found that second-order optimizers (SOOs) generally exhibited significantly lower variance at specific batch sizes, suggesting they may require less hyperparameter tuning, leading to a reduced overall time to solution for model training.