摘要
对覆盖区下伏岩体的有效识别是实现深部找矿突破的关键,近年来机器学习理论的发展为岩性识别提供了新的思路。梯度提升决策树(GBDT)算法是以决策树为基函数的集成学习方法,算法通过将学习得到的多个树模型进行集成,可以达到同时减少模型方差和偏差的效果。本文以地球化学元素数据为基础,利用GBDT算法进行岩性识别研究,并将所得结果与KNN、SVM和决策树3种机器学习算法进行对比,结果表明,针对岩性识别问题,GBDT算法具有更高的精度,可以作为岩性识别的参考技术,具有一定的推广和应用价值。
The effective identification of the concealed intrusion in the covered area is the key issue for deep mining. The development of Machine Learning theory provides a new way for lithology identification. The Gradient Boosting Decision Tree(GBDT) algorithm is an Ensemble Learning method based on the decision tree. Based on the geochemical element data, this paper used the GBDT algorithm to identify the lithology. In this study, we compare three traditional machine learning algorithms: KNN, SVM and Decision Tree. The results show that the GBDT algorithm has higher precision compared with the traditional machine learning algorithm, and it can be used as a technique for lithology identification.
作者
韩启迪
张小桐
申维
HAN Qi-di;ZHANG Xiao-tong;SHEN Wei(School of Earth Sciences and Resources China University of Geosciences (Beifing ),Beijing 100083,China;China Land Surveying and Planning Institute ,Beijing 100035,China)
出处
《矿物岩石地球化学通报》
CAS
CSCD
北大核心
2018年第6期1173-1180,共8页
Bulletin of Mineralogy, Petrology and Geochemistry
基金
国家自然科学基金项目(41172302,40672196)
关键词
机器学习
GBDT
岩性分类
决策树
machine learning
GBDT
lithology classification
Decision Tree
