DeepAI
Log In Sign Up

Training Deep Networks without Learning Rates Through Coin Betting

05/22/2017
by   Francesco Orabona, et al.
0

Deep learning methods achieve state-of-the-art performance in many application scenarios. Yet, these methods require a significant amount of hyperparameters tuning in order to achieve the best results. In particular, tuning the learning rates in the stochastic optimization process is still one of the main bottlenecks. In this paper, we propose a new stochastic gradient descent procedure for deep networks that does not require any learning rate setting. Contrary to previous methods, we do not adapt the learning rates nor we make use of the assumed curvature of the objective function. Instead, we reduce the optimization process to a game of betting on a coin and propose a learning-rate-free optimal algorithm for this scenario. Theoretical convergence is proven for convex and quasi-convex functions and empirical evidence shows the advantage of our algorithm over popular stochastic gradient algorithms.

READ FULL TEXT

page 1

page 2

page 3

page 4

02/15/2015

Equilibrated adaptive learning rates for non-convex optimization

Parameter-specific adaptive learning rate methods are computationally ef...
02/12/2020

LaProp: a Better Way to Combine Momentum with Adaptive Gradient

Identifying a divergence problem in Adam, we propose a new optimizer, La...
10/11/2022

Factors of Influence of the Overestimation Bias of Q-Learning

We study whether the learning rate α, the discount factor γ and the rewa...
02/22/2021

A Probabilistically Motivated Learning Rate Adaptation for Stochastic Optimization

Machine learning practitioners invest significant manual and computation...
10/18/2019

Robust Learning Rate Selection for Stochastic Optimization via Splitting Diagnostic

This paper proposes SplitSGD, a new stochastic optimization algorithm wi...
07/19/2019

Surfing: Iterative optimization over incrementally trained deep networks

We investigate a sequential optimization procedure to minimize the empir...

Code Repositories

cocob

TensorFlow implementation of COCOB


view repo

chainer-cocob

COCOB-Backprop (https://arxiv.org/abs/1705.07795) implementation for Chainer


view repo