An Introduction to Distributed Hybrid Hyperparameter Optimization- Jun Liu | SciPy 2022

Hyperparameter optimization in machine learning is commonly done on single search spaces, where the same search method is applied to all parameters. We introduce the concept of hybrid search space to combine various existing tuning methods in one optimization task through Fugue-tune. We will also demo how it simplifies the usage of Optuna and scales out on Spark.