To obtain access to full text of journal and articles you must register!
- Article name
- A METHOD FOR AUTOMATICALLY TUNING HYPERPARAMETERS OF A HYBRID DATA PREPARATION ALGORITHM BASED ON BAYESIAN OPTIMIZATION WITH ADAPTIVE WEIGHTING
- Authors
- Prudnikov S. I., , prudnikovscience@gmail.com, St. Petersburg Federal Research Center of the Russian Academy of Sciences, St. Petersburg, Russia
- Keywords
- Bayesian optimization / hyperparameters / Gaussian processes / adaptive weighting / data preparation / AutoML
- Year
- 2026 Issue 3 Pages 21 - 26
- Code EDN
- JRHUXI
- Code DOI
- 10.52190/2073-2597_2026_3_21
- Abstract
- This article presents a method for automatically tuning hyperparameters of hybrid data preparation algorithms for intelligent decision support systems. Unlike classical Bayesian optimization, the proposed approach includes: an adaptive mechanism for dynamically recalculating the weights of the objective function components; a hybrid Gaussian process kernel that takes into account the heterogeneity of hyperparameters; and warm initialization of the initial sample based on meta-heuristics. The method was experimentally compared with Random Search, Hyperband, BOHB, Optuna, and classical Bayesian optimization on three real-world datasets. A computational experiment demonstrated an improvement in the F1 score by 0.8-1.9 % and a reduction in computation time of up to 31 % relative to alternative methods.
- Text
- BUY for read the full text of article
- Buy
- 800.00 rub
