To Raise or Not To Raise: The Autonomous Learning Rate Question

TL;DR

This paper introduces an Autonomous Learning Rate Controller that dynamically adjusts the learning rate during training, reducing the need for manual tuning and improving robustness across different models and datasets.

Contribution

The paper presents a novel autonomous controller for learning rate adjustment, addressing the challenge of manual tuning and sensitivity to training conditions.

Findings

Reduces manual effort in selecting learning rates

Improves training robustness across architectures and datasets

Achieves competitive or better performance with adaptive learning rates

Abstract

There is a parameter ubiquitous throughout the deep learning world: learning rate. There is likewise a ubiquitous question: what should that learning rate be? The true answer to this question is often tedious and time consuming to obtain, and a great deal of arcane knowledge has accumulated in recent years over how to pick and modify learning rates to achieve optimal training performance. Moreover, the long hours spent carefully crafting the perfect learning rate can come to nothing the moment your network architecture, optimizer, dataset, or initial conditions change ever so slightly. But it need not be this way. We propose a new answer to the great learning rate question: the Autonomous Learning Rate Controller. Find it at https://github.com/fastestimator/ARC/tree/v2.0