Simultaneous Continuous/ Discrete Hyperparameter tuning with Policy Gradients

towards-data-science

This post was originally published by Frank Odom at Towards Data Science

We demonstrate an efficient method for simultaneously tuning discrete and continuous hyperparameters for machine learning models using policy gradients.

Overview

Review of Policy Gradients

Extending to Continuous Hyperparameters

Toy Problem: Continuous Hyperparameters

Average MAE: 2.155

Another Toy Problem: Mixed Hyperparameters

Average MAE: 2.088

Takeaways

Spread the word

This post was originally published by Frank Odom at Towards Data Science

Related posts