Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

OptunaHub: A Platform for Black-Box Optimization

Yoshihiko Ozaki, Shuhei Watanabe, Toshihiko Yanase

TL;DR

Black-box optimization research is fragmented across AutoML, Materials Informatics, and related fields, limiting cross-domain reuse of methods and benchmarks. OptunaHub proposes a centralized platform consisting of a unified Python API (OptunaHub Module), a contributor registry (OptunaHub Registry), and a web catalog (OptunaHub Web) to streamline discovery and evaluation of BBO methods. The Module provides load_module and base classes to enable seamless integration with the Optuna interface, while the Registry aggregates hundreds of contributed samplers, benchmarks, and tools, with thousands of downloads demonstrating reach. The Web interface auto-generates package pages with full-text and tag search, improving visibility and reusability of community contributions. Collectively, OptunaHub aims to accelerate BBO research by fostering a virtuous cycle of contributions and practical applications.

Abstract

Black-box optimization (BBO) drives advances in domains such as AutoML and Materials Informatics, yet research efforts often remain fragmented across domains. We introduce OptunaHub (https://hub.optuna.org/), a community platform that centralizes BBO methods and benchmarks. OptunaHub provides unified Python APIs, a contributor package registry, and a web interface to promote searchability and cross-domain research. OptunaHub aims to foster a virtuous cycle of contributions and applications. The source code is publicly available in the optunahub, optunahub-registry, and optunahub-web repositories under the Optuna organization on GitHub (https://github.com/optuna/).

OptunaHub: A Platform for Black-Box Optimization

TL;DR

Black-box optimization research is fragmented across AutoML, Materials Informatics, and related fields, limiting cross-domain reuse of methods and benchmarks. OptunaHub proposes a centralized platform consisting of a unified Python API (OptunaHub Module), a contributor registry (OptunaHub Registry), and a web catalog (OptunaHub Web) to streamline discovery and evaluation of BBO methods. The Module provides load_module and base classes to enable seamless integration with the Optuna interface, while the Registry aggregates hundreds of contributed samplers, benchmarks, and tools, with thousands of downloads demonstrating reach. The Web interface auto-generates package pages with full-text and tag search, improving visibility and reusability of community contributions. Collectively, OptunaHub aims to accelerate BBO research by fostering a virtuous cycle of contributions and practical applications.

Abstract

Black-box optimization (BBO) drives advances in domains such as AutoML and Materials Informatics, yet research efforts often remain fragmented across domains. We introduce OptunaHub (https://hub.optuna.org/), a community platform that centralizes BBO methods and benchmarks. OptunaHub provides unified Python APIs, a contributor package registry, and a web interface to promote searchability and cross-domain research. OptunaHub aims to foster a virtuous cycle of contributions and applications. The source code is publicly available in the optunahub, optunahub-registry, and optunahub-web repositories under the Optuna organization on GitHub (https://github.com/optuna/).

Paper Structure

This paper contains 6 sections, 2 figures.

Table of Contents

  1. Introduction
  2. OptunaHub Ecosystem
  3. OptunaHub Module: Unified Easy-to-Use APIs Compatible with Optuna
  4. OptunaHub Registry: Gateway for Contributions from Community
  5. OptunaHub Web: Package Catalog with Full-Text & Tag Search
  6. Final Remarks

Figures (2)

  • Figure 1: The conceptual visualization of the relationship between each OptunaHub component. Essentially, OptunaHub Module allows users to load packages in OptunaHub Registry, and the API documentation of each package is available at OptunaHub Web. All registered packages can be seamlessly integrated with the Optuna interface, making it possible to perform BBO with different samplers on different problems with minimal modifications. Additionally, OptunaHub Web is equipped with full-text search, helping users search for packages efficiently.
  • Figure 2: Package download (load_module) statistics of OptunaHub. Left: Top-5 most downloaded packages in OptunaHub Registry in September 2025. https://hub.optuna.org/samplers/auto_sampler/ and https://hub.optuna.org/samplers/nsgaii_with_initial_trials/ are methods implemented by the Optuna team to meet practical demands. We defer the details of https://hub.optuna.org/samplers/tpe_tutorial/, https://hub.optuna.org/samplers/catcma/, and https://hub.optuna.org/samplers/implicit_natural_gradient/ to watanabe2023tree, hamano2024catcma, and lyu2019black, respectively. Right: Monthly total package downloads over time. The monthly downloads have steadily grown since the OptunaHub's beta version release in July 2024, increasing the visibility of registered packages among the community.