A/B testing while experimenting with Machine Learning models
Get user feedback by deploying many models and by trying A/B testing
The growth in robotics technology is greatly intertwined with the growth in the technology of Machine Learning, because robots need to learn to contribute something useful. In this forum, we have explored the use of Machine Learning in robotics in several ways… one of the main ways has been the exercise of training one Vector robot to detect another Vector robot. In that exercise, we investigated how to build a Machine Learning (ML) model which could accomplish the task of detecting a Vector robot in a image coming from Vector’s camera feed. In this article, we will discuss how to improve the model using a technology known as A/B testing.
A/B testing is extremely important while evaluating different ML models. In this article, we try to understand what A/B testing is, and why it is specifically important in the context of deploying a Machine Learning model for usage in production. Let us first investigate the need to do A/B testing in the first place.