On the Convergence of Adam and Beyond

By Sashank J. Reddi, Satyen Kale, Surinder Kumar. 2018

2,997 citations recorded by the source corpus.

Several recently proposed stochastic optimization methods that have been successfully used in training deep networks such as RMSProp, Adam, Adadelta, Nadam are based on using gradient updates scaled by square roots of exponential moving averages of squared past gradients. In many applications, e.g. learning with large output spaces, it has been empirically observed that these algorithms fail to converge to an optimal solution (or a critical point in nonconvex settings). We show that one cause for such failures is the exponential moving average used in the algorithms. We provide an explicit example of a simple convex optimization setting where Adam does not converge to the optimal solution, and describe the precise problems with the previous analysis of Adam algorithm. Our analysis suggests that the convergence issues can be fixed by endowing such algorithms with `long-term memory' of past gradients, and propose new variants of the Adam algorithm which not only fix the convergence issues but often also lead to improved empirical performance.

Subfields: Artificial Intelligence, Numerical Analysis, Optimization & Training, Philosophy

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

  1. Mohit Bansal
  2. Yoshua Bengio
  3. Jianfeng Gao
  4. Percy Liang
  5. Quoc V. Le
  6. Luke S. Zettlemoyer
  7. Sergey Levine
  8. Ruslan Salakhutdinov

In Numerical Analysis

  1. Simon Lacoste-Julien
  2. Nicolas Loizou
  3. Jakub Konecný
  4. Sebastian Pokutta
  5. Francesco Orabona
  6. Sebastian U. Stich
  7. Mikhail Belkin
  8. Issam Hadj Laradji

In Optimization & Training

  1. Simon Lacoste-Julien
  2. Peter Richtárik
  3. Martin Jaggi
  4. Nicolas Loizou
  5. Sebastian U. Stich
  6. Francis R. Bach
  7. Jakub Konecný
  8. Quoc V. Le

In Philosophy

  1. Mikhail Belkin
  2. Chaoyue Liu
  3. Jean-Marc Valin
  4. Diganta Misra
  5. Jan Skoglund
  6. Francesco Orabona
  7. Yang Song
  8. Stefano Ermon

Citations in the corpus