On the Job Training

TL;DR

This paper introduces a new framework for developing and assessing machine learning agents capable of continuous learning and adaptation in real-world scenarios, emphasizing the importance of system design on performance.

Contribution

It presents a novel framework for building lifelong learning agents, along with evaluation metrics and insights into how system construction impacts performance.

Findings

Subtle system design changes greatly influence agent effectiveness.

The proposed metrics enable better assessment of lifelong learning capabilities.

Framework supports agents that learn continuously during their operational life.

Abstract

We propose a new framework for building and evaluating machine learning algorithms. We argue that many real-world problems require an agent which must quickly learn to respond to demands, yet can continue to perform and respond to new training throughout its useful life. We give a framework for how such agents can be built, describe several metrics for evaluating them, and show that subtle changes in system construction can significantly affect agent performance.