Optuna

Optuna
DeveloperPreferred Networks
ReleaseVersion 1.0 (January, 2020)
Stable release
Version 4.4.0 (June 16, 2025)
Written inPython
Available inEnglish
LicenseMIT License
Websiteoptuna.org
Repositorygithub.com/optuna/optuna

Optuna is an open-source Python library for automatic hyperparameter tuning of machine learning models. It was first introduced in 2018 by Preferred Networks, a Japanese startup that works on practical applications of deep learning in various fields. The beta version of Optuna was released at the end of the year, with the subsequent first major stable release announced in January 2020.

Hyperparameter optimization

Hyperparameter Optimization using Grid Search

Hyperparameter optimization involves finding the optimal value of non-trainable parameters, defined by the user. Examples of hyperparameters are learning rate, number of layers or neurons, regularization strength, and tree depth. However, the optimal values strongly depend on the specific algorithm (e.g., classification, regression, or clustering).

Hyperparameter optimization can be especially relevant when dealing with large-scale problems or limited resources, as it improves model accuracy, reduces overfitting, and decreases training time. However, when the hyperparameter space increases, it may become computationally expensive. Hence, there are methods (e.g., grid search, random search, or Bayesian optimization) that considerably simplify this process.

Features

Optuna is designed to optimize model hyperparameters by searching large spaces and discarding combinations that show no significant improvement in the model. Moreover, it can parallelize the hyperparameter search over multiple threads or processes. Optuna works with a high degree of modularity, allowing the definition of different configurations for searching hyperparameters. It is possible to choose which hyperparameters to optimize and how to optimize them. Additionally, it permits parameter customization at runtime, meaning that the values to explore for the hyperparameters (i.e., the search space) can be defined while writing the code, rather than being defined in advance.

Hyperparameter Optimization using Tree-Structured Parzen Estimators

Sampling

Optuna exploits well-established algorithms to perform hyperparameter optimization, progressively reducing the search space in light of objective values. Examples include Gaussian-process-based algorithms (i.e., a Gaussian process to model the objective function), tree-structured Parzen estimator (TPE) (i.e., a model-based optimization method that estimates the objective function and selects the best hyperparameters), and random search (i.e., a basic optimization approach used for benchmarking).

Early stopping

Bayesian optimization of a function (black) with a gaussian process (purple). Three acquisition functions (blue).

Optuna includes a pruning feature to stop trials early when the results show no significant improvement in the model. This prevents unnecessary computation and is used for models with long training times to save time and computational resources. Specifically, Optuna exploits techniques such as median and threshold pruning.

Scalability

Optuna is designed to scale with distributed computing resources, supporting parallel execution. This feature allows users to run optimization trials across multiple processors or machines.

Integration with third-party libraries

Optuna integrates with various machine-learning (ML) libraries and frameworks:

Moreover, Optuna offers a real-time dashboard that allows a user to monitor the optimization history and the hyperparameter importance through graphs and tables.

Applications

Optuna was designed to be framework-agnostic so that it can be used with any machine-learning (ML) or deep-learning (DL) framework.

Standard pipeline for the optimization of hyperparameters for the training of a Decision Tree Classifier using Optuna

Machine learning

Optuna can be used to optimize the hyperparameters of ML models. Examples include:

Deep learning

In the context of deep learning, Optuna can be used during the training of neural networks (NN) to optimize the learning rate, batch size, and number of hidden layers. For example, it can be used for:

Domains

Optuna has found applications in various research studies and industry implementations across different domains.

Healthcare

In healthcare, Optuna is currently used for medical image analysis to optimize DL models for tumor detection, disease identification, multi-organ semantic segmentation, and radiological image analysis. It can also be used for disease prediction, in particular to improve the accuracy of predictive models for disease diagnosis and treatment outcomes. It can also be used in genomics to predict genetic diseases and identify genetic variations.

Finance

In finance, Optuna is used to optimize models for algorithmic trading. It can be used to predict market movements because of its ability to handle wide parameter ranges and complex model structures. In particular, it is used for financial risk analysis and forecasting. The risks addressed are credit, market, and operational risks.

Autonomous systems

Optuna is used for real-time applications in autonomous systems for robotics, supporting decision-making in dynamic environments. It is also used in the context of self-driving cars to optimize models for navigating safely in complex environments. For example, Optuna can be used in scenarios where there is a need to evaluate the rate and severity of accidents or address network intrusion attacks due to potential vulnerabilities.

Natural language processing (NLP)

Optuna has also been applied in NLP. For example, it has been used for text classification to classify written text (e.g., spam vs. not spam, topic modeling). Another task is sentiment analysis, namely the detection of feelings expressed in text, which is particularly used to analyze content from social media and customer reviews.

Reinforcement learning (RL)

Regarding RL, Optuna is used in gaming to improve model performance in games and virtual environments. In robotics, it has been used to optimize decision-making processes in robotic systems for tasks such as manipulation and navigation. Additionally, in the field of autonomous vehicles, Optuna can optimize RL models to improve safety and navigation efficiency.

See also

References

  1. 1 2 3 4 5 6 7 8 Akiba, Takuya; Sano, Shotaro; Yanase, Toshihiko; Ohta, Takeru; Koyama, Masanori (2019-07-25). "Optuna: A Next-generation Hyperparameter Optimization Framework". Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. KDD '19. New York, NY, USA: Association for Computing Machinery. pp. 2623–2631. doi:10.1145/3292500.3330701. ISBN 978-1-4503-6201-6.
  2. 1 2 Wang, Zhaofei; Li, Hao; Wang, Qiuping (2025). "An Approach to Truck Driving Risk Identification: A Machine Learning Method Based on Optuna Optimization". IEEE Access. 13: 42723–42732. Bibcode:2025IEEEA..1342723W. doi:10.1109/ACCESS.2025.3547445. ISSN 2169-3536.
  3. 1 2 Li, Yifan; Cao, Yanpeng; Yang, Jintang; Wu, Mingyu; Yang, Aimin; Li, Jie (2024-06-01). "Optuna-DFNN: An Optuna framework driven deep fuzzy neural network for predicting sintering performance in big data". Alexandria Engineering Journal. 97: 100–113. doi:10.1016/j.aej.2024.04.026. ISSN 1110-0168.
  4. 1 2 Pinichka, Chayut; Chotpantarat, Srilert; Cho, Kyung Hwa; Siriwong, Wattasit (2025-07-01). "Comparative analysis of SWAT and SWAT coupled with XGBoost model using Optuna hyperparameter optimization for nutrient simulation: A case study in the Upper Nan River basin, Thailand". Journal of Environmental Management. 388 126053. Bibcode:2025JEnvM.38826053P. doi:10.1016/j.jenvman.2025.126053. ISSN 0301-4797. PMID 40460531.
  5. 1 2 3 Huang, Jiancong; Rojas, Juan; Zimmer, Matthieu; Wu, Hongmin; Guan, Yisheng; Weng, Paul (2021-03-08). "Hyperparameter Auto-Tuning in Self-Supervised Robotic Learning". IEEE Robotics and Automation Letters. 6 (2): 3537–3544. arXiv:2010.08252. Bibcode:2021IRAL....6.3537H. doi:10.1109/LRA.2021.3064509. ISSN 2377-3766.
  6. ↑ Yang, Li; Shami, Abdallah (2020-11-20). "On hyperparameter optimization of machine learning algorithms: Theory and practice". Neurocomputing. 415: 295–316. arXiv:2007.15745. doi:10.1016/j.neucom.2020.07.061. ISSN 0925-2312.
  7. ↑ Tanim, Ahad Hasan; Smith-Lewis, Corinne; Downey, Austin R. J.; Imran, Jasim; Goharian, Erfan (2024-08-01). "Bayes_Opt-SWMM: A Gaussian process-based Bayesian optimization tool for real-time flood modeling with SWMM". Environmental Modelling & Software. 179 106122. Bibcode:2024EnvMS.17906122T. doi:10.1016/j.envsoft.2024.106122. ISSN 1364-8152. OSTI 2378088.
  8. ↑ Ozaki, Yoshihiko; Tanigaki, Yuki; Watanabe, Shuhei; Nomura, Masahiro; Onishi, Masaki (2022-04-08). "Multiobjective Tree-Structured Parzen Estimator". Journal of Artificial Intelligence Research. 73: 1209–1250. doi:10.1613/jair.1.13188. ISSN 1076-9757.
  9. ↑ Bergstra, James; Bengio, Yoshua (2012-02-01). "Random search for hyper-parameter optimization". J. Mach. Learn. Res. 13: 281–305. ISSN 1532-4435.
  10. ↑ Wistuba, Martin; Schilling, Nicolas; Schmidt-Thieme, Lars (2015). "Hyperparameter Search Space Pruning – A New Component for Sequential Model-Based Hyperparameter Optimization". In Appice, Annalisa; Rodrigues, Pedro Pereira; Santos Costa, Vítor; Gama, João; Jorge, Alípio; Soares, Carlos (eds.). Machine Learning and Knowledge Discovery in Databases. Lecture Notes in Computer Science. Vol. 9285. Cham: Springer International Publishing. pp. 104–119. doi:10.1007/978-3-319-23525-7_7. ISBN 978-3-319-23525-7.
  11. ↑ Parra-Ullauri, Juan; Zhang, Xunzheng; Bravalheri, Anderson; Nejabati, Reza; Simeonidou, Dimitra (2023-06-07). "Federated Hyperparameter Optimisation with Flower and Optuna". Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing. SAC '23. New York, NY, USA: Association for Computing Machinery. pp. 1209–1216. doi:10.1145/3555776.3577847. ISBN 978-1-4503-9517-5.
  12. ↑ Almarzooq, Hussain; Waheed, Umair bin (2024-05-21). "Automating hyperparameter optimization in geophysics with Optuna: A comparative study". Geophysical Prospecting. 72 (5): 1778–1788. Bibcode:2024GeopP..72.1778A. doi:10.1111/1365-2478.13484. ISSN 1365-2478.
  13. ↑ Rokach, Lior; Maimon, Oded (2005), "Decision Trees", in Maimon, Oded; Rokach, Lior (eds.), Data Mining and Knowledge Discovery Handbook, Boston, MA: Springer US, pp. 165–192, doi:10.1007/0-387-25465-x_9, ISBN 978-0-387-25465-4
  14. ↑ R, Shyam; Ayachit, Sai Sanjay; Patil, Vinayak; Singh, Anubhav (2020-12-18). "Competitive Analysis of the Top Gradient Boosting Machine Learning Algorithms". 2020 2nd International Conference on Advances in Computing, Communication Control and Networking (ICACCCN). pp. 191–196. doi:10.1109/ICACCCN51052.2020.9362840. ISBN 978-1-7281-8337-4.
  15. ↑ Pisner, Derek A.; Schnyer, David M. (2020-01-01), "Chapter 6 - Support vector machine", in Mechelli, Andrea; Vieira, Sandra (eds.), Machine Learning, Academic Press, pp. 101–121, ISBN 978-0-12-815739-8, retrieved 2025-07-08
  16. ↑ Zhang, Zhongheng (2016-06-14). "Introduction to machine learning: k-nearest neighbors". Annals of Translational Medicine. 4 (11): 218. doi:10.21037/atm.2016.03.37. ISSN 2305-5839. PMC 4916348. PMID 27386492.
  17. ↑ Tripepi, G.; Jager, K. J.; Dekker, F. W.; Zoccali, C. (2008-04-01). "Linear and logistic regression analysis". Kidney International. 73 (7): 806–810. doi:10.1038/sj.ki.5002787. ISSN 0085-2538. PMID 18200004.
  18. ↑ Yang, Feng-Jen (2018-12-12). "An Implementation of Naive Bayes Classifier". 2018 International Conference on Computational Science and Computational Intelligence (CSCI). pp. 301–306. doi:10.1109/CSCI46756.2018.00065. ISBN 978-1-7281-1360-9.
  19. ↑ Li, Zewen; Liu, Fan; Yang, Wenjie; Peng, Shouheng; Zhou, Jun (2021-06-10). "A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects". IEEE Transactions on Neural Networks and Learning Systems. 33 (12): 6999–7019. arXiv:2004.02806. doi:10.1109/TNNLS.2021.3084827. hdl:10072/405164. ISSN 2162-2388. PMID 34111009.
  20. ↑ Caterini, Anthony L.; Chang, Dong Eui (2018-03-23), "Recurrent Neural Networks", in Caterini, Anthony L.; Chang, Dong Eui (eds.), Deep Neural Networks in a Mathematical Framework, Cham: Springer International Publishing, pp. 59–79, doi:10.1007/978-3-319-75304-1_5, ISBN 978-3-319-75304-1
  21. ↑ Gillioz, Anthony; Casas, Jacky; Mugellini, Elena; Khaled, Omar Abou (2020-09-06). "Overview of the Transformer-based Models for NLP Tasks". 2020 15th Conference on Computer Science and Information Systems (FedCSIS). Proceedings of the 2020 Federated Conference on Computer Science and Information Systems. 21: 179–183. doi:10.15439/2020F20. ISBN 978-83-955416-7-4.{{cite journal}}: CS1 maint: periodical has ISBN (link)
  22. ↑ Bhuvanya, R.; T.Kujani; Kumaran, S.Manoj; Lokesh Kumar, N. (2024-11-22). "OptNet: Innovative Model for Early Lung Cancer Diagnosis integrating TabNet and Optuna". 2024 International Conference on Electronic Systems and Intelligent Computing (ICESIC). pp. 174–179. doi:10.1109/ICESIC61777.2024.10846378. ISBN 979-8-3315-2298-8.
  23. ↑ Kumar Sahu, Prabhat; Fatma, Taiyaba (2025-02-07). "Optimized Breast Cancer Classification Using PCA-LASSO Feature Selection and Ensemble Learning Strategies With Optuna Optimization". IEEE Access. 13: 35645–35661. Bibcode:2025IEEEA..1335645K. doi:10.1109/ACCESS.2025.3539746. ISSN 2169-3536.
  24. ↑ Deepan, P.; Prabhakar Reddy, G.; Arsha Reddy, M.; Vidya, R.; Dhiravidaselvi, S. (2024), "Maximizing Accuracy in Alzheimer's Disease Prediction: A Optuna Hyper Parameter Optimization Strategy Using MRI Images", in Bhattacharya, Pronaya; Liu, Haipeng; Dutta, Pushan Kumar; Rodrigues, Joel J. P. C. (eds.), Revolutionizing Healthcare 5.0: The Power of Generative AI: Advancements in Patient Care Through Generative AI Algorithms, Cham: Springer Nature Switzerland, pp. 77–91, doi:10.1007/978-3-031-75771-6_5, ISBN 978-3-031-75771-6
  25. ↑ Vaiyapuri, Thavavel (2025-08-01). "An Optuna-Based Metaheuristic Optimization Framework for Biomedical Image Analysis". Engineering, Technology & Applied Science Research. 15 (4): 24382–24389. doi:10.48084/etasr.11234. ISSN 1792-8036.
  26. ↑ Srinivas, Polipireddy; Katarya, Rahul (2022-03-01). "hyOPTXg: OPTUNA hyper-parameter optimization framework for predicting cardiovascular disease using XGBoost". Biomedical Signal Processing and Control. 73 103456. doi:10.1016/j.bspc.2021.103456. ISSN 1746-8094.
  27. ↑ Shen, Zijie; Shen, Enhui; Zhu, Qian-Hao; Fan, Longjiang; Zou, Quan; Ye, Chu-Yu (2023). "GSCtool: A Novel Descriptor that Characterizes the Genome for Applying Machine Learning in Genomics". Advanced Intelligent Systems. 5 (12) 2300426. doi:10.1002/aisy.202300426. ISSN 2640-4567.
  28. ↑ Shui, Hongyi; Sha, Xinye; Chen, Baizheng; Wu, Jiajie (2024-08-26). "Stock weighted average price prediction based on feature engineering and Lightgbm model". Proceedings of the 2024 International Conference on Digital Society and Artificial Intelligence. DSAI '24. New York, NY, USA: Association for Computing Machinery. pp. 336–340. doi:10.1145/3677892.3677945. ISBN 979-8-4007-0983-8.
  29. 1 2 Garg, Deepak; Shelke, Nitin Arvind; Kitukale, Gauri; Mehlawat, Nishka (2024-04-05). "Leveraging Financial Data and Risk Management in Banking sector using Machine Learning". 2024 IEEE 9th International Conference for Convergence in Technology (I2CT). pp. 1–6. doi:10.1109/I2CT61223.2024.10544336. ISBN 979-8-3503-9445-0.
  30. ↑ Rezashoar, Soheil; Kashi, Ehsan; Saeidi, Soheila (2024-07-26). "A hybrid algorithm based on machine learning (LightGBM-Optuna) for road accident severity classification (case study: United States from 2016 to 2020)". Innovative Infrastructure Solutions. 9 (8): 319. Bibcode:2024InnIS...9..319R. doi:10.1007/s41062-024-01626-y. ISSN 2364-4184.
  31. ↑ Jha, Jayesh; Yadav, Jatin; Naqvi, Haider (2025). "MILCCDE: A Metaheuristic Improved Decision-Based Ensemble Framework for Intrusion Detection in Autonomous Vehicles". In Bansal, Jagdish Chand; Jamwal, Prashant K.; Hussain, Shahid (eds.). Sustainable Computing and Intelligent Systems. Lecture Notes in Networks and Systems. Vol. 1295. Singapore: Springer Nature. pp. 255–267. doi:10.1007/978-981-96-3311-1_21. ISBN 978-981-96-3311-1.
  32. ↑ Parekh, Nishank; Sen, Arzob; Rajasekaran, P.; Jayaseeli, J. D. Dorathi; Robert, P. (2024-12-17). "Network Intrusion Detection System Using Optuna". 2024 International Conference on IoT Based Control Networks and Intelligent Systems (ICICNIS). pp. 312–318. doi:10.1109/ICICNIS64247.2024.10823298. ISBN 979-8-3315-1809-7.
  33. ↑ Rahmi, Nadya Alinda; Defit, Sarjon; Okfalisa (2024-12-31). "The Use of Hyperparameter Tuning in Model Classification: A Scientific Work Area Identification". International Journal on Informatics Visualization. 8 (4): 2181. doi:10.62527/joiv.8.4.3092. ISSN 2549-9904.
  34. ↑ Efendi, Akmar; Fitri, Iskandar; Nurcahyo, Gunadi Widi (2024-08-07). "Improvement of Machine Learning Algorithms with Hyperparameter Tuning on Various Datasets". 2024 International Conference on Future Technologies for Smart Society (ICFTSS). pp. 75–79. doi:10.1109/ICFTSS61109.2024.10691354. ISBN 979-8-3503-7384-4.
  35. ↑ Kiran, B Ravi; Sobh, Ibrahim; Talpaert, Victor; Mannion, Patrick; Sallab, Ahmad A. Al; Yogamani, Senthil; Pérez, Patrick (2021-02-09). "Deep Reinforcement Learning for Autonomous Driving: A Survey". IEEE Transactions on Intelligent Transportation Systems. 23 (6): 4909–4926. arXiv:2002.00444. Bibcode:2022ITITr..23.4909K. doi:10.1109/TITS.2021.3054625. ISSN 1558-0016.

Further reading