Timezone: »

Adam with Bandit Sampling for Deep Learning
Rui Liu · Tianyi Wu · Barzan Mozafari

Wed Dec 09 09:00 AM -- 11:00 AM (PST) @ Poster Session 3 #811
Adam is a widely used optimization method for training deep learning models. It computes individual adaptive learning rates for different parameters. In this paper, we propose a generalization of Adam, called Adambs, that allows us to also adapt to different training examples based on their importance in the model's convergence. To achieve this, we maintain a distribution over all examples, selecting a mini-batch in each iteration by sampling according to this distribution, which we update using a multi-armed bandit algorithm. This ensures that examples that are more beneficial to the model training are sampled with higher probabilities. We theoretically show that Adambs improves the convergence rate of Adam---$O(\sqrt{\frac{\log n}{T} })$ instead of $O(\sqrt{\frac{n}{T}})$ in some cases. Experiments on various models and datasets demonstrate Adambs's fast convergence in practice.

Author Information

Rui Liu (University of Michigan, Ann Arbor)
Tianyi Wu (University of Michigan, Ann Arbor)
Barzan Mozafari (University of Michigan)

Related Events (a corresponding poster, oral, or spotlight)