Timezone: »

Stochastic Polyak Stepsize with a Moving Target
Robert Gower · Aaron Defazio · Mike Rabbat

We propose a new stochastic gradient method that uses recorded past loss values to compute adaptive stepsizes. Our starting point is to show that the SP (Stochastic Polyak) method directly exploits interpolated models. That is, SP is a subsampled Newton-Raphson method applied to solving certain interpolation equations. These interpolation equations only hold for models that interpolate the data. We then use this viewpoint to develop a new variant of the SP method that converges without interpolation called MOTAPS. The MOTAPS method uses n auxiliary variables, one for each data point, that track the loss value for each data point. These auxiliary variables and the loss values are then used to set the step size.

We provide a global convergence theory for MOTAPS by showing that it can be interpreted as a special variant of online SGD. We also perform several numerical experiments on convex learning problems, and non-convex learning problem based on image classification and language translation. In all of our tasks we show that MOTAPS is competitive with the relevant baseline method.

Author Information

Robert Gower (Flatiron Institute)
Aaron Defazio (Facebook AI Research)
Mike Rabbat (Facebook FAIR)

More from the Same Authors