Timezone: »

 
Poster
A new regret analysis for Adam-type algorithms
Ahmet Alacaoglu · Yura Malitsky · Panayotis Mertikopoulos · Volkan Cevher

Wed Jul 15 12:00 PM -- 12:45 PM & Thu Jul 16 12:00 AM -- 12:45 AM (PDT) @
in Poster Session 26 »
In this paper, we focus on a theory-practice gap for Adam and its variants (AMSGrad, AdamNC, etc.). In practice, these algorithms are used with a constant first-order moment parameter $\beta_{1}$ (typically between $0.9$ and $0.99$). In theory, regret guarantees for online convex optimization require a rapidly decaying $\beta_{1}\to0$ schedule. We show that this is an artifact of the standard analysis, and we propose a novel framework that allows us to derive optimal, data-dependent regret bounds with a constant $\beta_{1}$, without further assumptions. We also demonstrate the flexibility of our analysis on a wide range of different algorithms and settings.
Keywords: Convex Optimization · Optimization · Optimization - Convex

Author Information

Ahmet Alacaoglu (EPFL)
Yura Malitsky (EPFL)
Panayotis Mertikopoulos (CNRS and Criteo AI Lab)
Volkan Cevher (EPFL)

More from the Same Authors