The fastly developing hyperparameter tuning framework Optuna has gained much attention in the machine learning community due to its high flexibility and rich features. One of the most important features in Optuna is the visualization module, which provides various APIs to help users to analyze the optimization process and possibly adjust or early stop the process correspondingly. However, currently, these APIs are called by the user manually and thus makes the monitoring process hard. In this project, we aim to develop a web dashboard for Optuna (Optuna-dashboard) to monitor various optimization processes, which will greatly simplify the monitoring process by automatically visualizing the optimization “study” in optuna in real-time.

Organization

Student

Huzi Cheng

Mentors

  • Crissman Loomis
  • Masashi Shibata
close

2021