Timezone: »

Optimizer Benchmarking Needs to Account for Hyperparameter Tuning
Prabhu Teja Sivaprasad · Florian Mai · Thijs Vogels · Martin Jaggi · François Fleuret

Thu Jul 16 12:00 PM -- 12:45 PM & Thu Jul 16 11:00 PM -- 11:45 PM (PDT) @ Virtual

The performance of optimizers, particularly in deep learning, depends considerably on their chosen hyperparameter configuration. The efficacy of optimizers is often studied under near-optimal problem-specific hyperparameters, and finding these settings may be prohibitively costly for practitioners. In this work, we argue that a fair assessment of optimizers' performance must take the computational cost of hyperparameter tuning into account, i.e., how easy it is to find good hyperparameter configurations using an automatic hyperparameter search. Evaluating a variety of optimizers on an extensive set of standard datasets and architectures, our results indicate that Adam is the most practical solution, particularly in low-budget scenarios.

Author Information

Prabhu Teja Sivaprasad (Idiap Research Institute)
Florian Mai (Idiap Research Institute)
Thijs Vogels (EPFL)
Martin Jaggi (EPFL)
François Fleuret (University of Geneva)

More from the Same Authors