基于Bayesian采样主动机器学习模型的6061铝合金成分精细优化

赵婉辰, 郑晨, 肖斌, 刘行, 刘璐, 余童昕, 刘艳洁, 董自强, 刘轶, 周策, 吴洪盛, 路宝坤

Composition Refinement of 6061 Aluminum Alloy Using Active Machine Learning Model Based on Bayesian Optimization Sampling

ZHAO Wanchen, ZHENG Chen, XIAO Bin, LIU Xing, LIU Lu, YU Tongxin, LIU Yanjie, DONG Ziqiang, LIU Yi, ZHOU Ce, WU Hongsheng, LU Baokun

图4 序列前向选择(SFS)与序列后向选择(SBS)特征选择方法与结果对比 (a) subset building results of SFS (RMSE—root mean square error, svr.rbf—radial basis function support vector regression) (b) subset filtering results of SBS (c) comparisons of modeling results between the two feature subsets (bpnn—back propagation neural network, svr.poly—polynomial kernel functions for support vector regression, gbr—gradient boosted regression, rfr—random forest regression)

Fig.4 Feature selection method and result comparisons between sequential forward selection (SFS) and sequential backward selection (SBS)