基于局部搜索贝叶斯算法的Xgboost参数选择

Parameter selection of xgboost based on local search bayesian algorithm

提出了一种基于密度的局部搜索贝叶斯算法的Xgboost参数选择方法(BOA-DLS-Xgboost).基于密度的局部搜索贝叶斯算法(BOA-DLS)在选择初始种群时采用拉丁超立方抽样(LHS),使初始种群更均匀地分布于参数空间;每次探索过程是在LHS抽样点的基础上,对稀疏点和当前最优解周围进行局部搜索得到待采样集,以提高解的收敛速度和精度.仿真实验结果表明:BOA-DLS比BOA具有更好的优化性能.利用BOA-DLS对Xgboost算法的参数进行优化,通过与四种经典集成学习算法以及BOA-Xgboost算法比较,所提出的BOA-DLS-Xgboost算法在参数优化方面的应用是合理有效的.

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.