We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于局部搜索贝叶斯算法的Xgboost参数选择.
- Authors
肖海军; 阚渟渟; 李春辉
- Abstract
A density-based local search Bayesian algorithm for Xgboost parameter selection method (BOA-DLS-Xgboost) is proposed. In the density-based local search Bayesian algorithm (BOA-DLS), Latin hypercube sampling (LHS) is used to select the initial population, which makes the initial population more uniformly distributed in the parameter space. In each exploration process, based on LHS sampling points, the local search is conducted around the sparse point and the current best point to obtain a new sample set, this process can improve the solution convergence speed and accuracy. The simulation experiments show BOA-DLS has better optimization performance than BOA. BOA-DLS are applied for optimizing the parameters of Xgboost algorithm, then compared with BOA-Xgboost and four ensemble learning algorithms, which show that the algorithms proposed are effective and efficient in the parameter optimization of Xgboost algorithm.
- Publication
Journal of South-Central Minzu University (Natural Science Edition), 2023, Vol 42, Issue 2, p201
- ISSN
1672-4321
- Publication type
Article
- DOI
10.20056/j.cnki.ZNMDZK.20230209