Measuring Competency of Machine Learning Systems and Enforcing Reliability

TL;DR

This paper investigates how environmental conditions affect machine learning system performance and proposes real-time competency assessments to enhance reliability, demonstrated through a self-driving vehicle simulation using visual imagery.

Contribution

It introduces a method to represent environmental conditions impacting ML agent strategies and performance, improving reliability through real-time competency evaluation.

Findings

Environmental conditions significantly influence ML agent performance.

Real-time competency assessments can maintain operator expectations.

The approach improves obstacle avoidance in simulated self-driving vehicles.

Abstract

We explore the impact of environmental conditions on the competency of machine learning agents and how real-time competency assessments improve the reliability of ML agents. We learn a representation of conditions which impact the strategies and performance of the ML agent enabling determination of actions the agent can make to maintain operator expectations in the case of a convolutional neural network that leverages visual imagery to aid in the obstacle avoidance task of a simulated self-driving vehicle.