This paper presents a hybrid ensemble classifier combined synthetic minority oversampling technique(SMOTE),random search(RS)hyper-parameters optimization algorithm and gradient boosting tree(GBT)to achieve efficient and accurate rock trace identification.A thirteen-dimensional database consist-ing of basic,vector,and discontinuity features is established from image samples.All data points are clas-sified as either"trace"or"non-trace"to divide the ultimate results into candidate trace samples.It is found that the SMOTE technology can effectively improve classification performance by recommending an optimized imbalance ratio of 1:5 to 1:4.Then,sixteen classifiers generated from four basic machine learning(ML)models are applied for performance comparison.The results reveal that the proposed RS-SMOTE-GBT classifier outperforms the other fifteen hybrid ML algorithms for both trace and non-trace classifications.Finally,discussions on feature importance,generalization ability and classification error are conducted for the proposed classifier.The experimental results indicate that more critical fea-tures affecting the trace classification are primarily from the discontinuity features.Besides,cleaning up the sedimentary pumice and reducing the area of fractured rock contribute to improving the overall clas-sification performance.The proposed method provides a new alternative approach for the identification of 3D rock trace.

Jiayao Chen;Hongwei Huang;Anthony G.Cohn;Dongming Zhang;Mingliang Zhou

Key Laboratory of Geotechnical and Underground Engineering of Ministry of Education,Department of Geotechnical Engineering,Tongji University,Shanghai 200092,China;School of Computing,University of Leeds,LS2 9JT Leeds,United Kingdom;Department of Computer Science and Technology,Tongji University,Shanghai 211985,China;School of Civil Engineering,Shandong University,Jinan 250061,China;Luzhong Institute of Safety,Environmental Protection Engineering and Materials,Qingdao University of Science and Technology,Zibo 255000,China;School of Mechanical and Electrical Engineering,Qingdao University of Science and Technology,Qingdao 260061,China

矿业科学技术学报（英文版）

[1]Jiayao Chen;Hongwei Huang;Anthony G.Cohn;Dongming Zhang;Mingliang Zhou-.Machine learning-based classification of rock discontinuity trace:SMOTE oversampling integrated with GBT ensemble learning)[J].矿业科学技术学报（英文版）,2022(02):309-322

pumice

Machine,learning,rock,discontinuity,trace,SMOTE,oversampling,integrated,GBT,ensemble,This,paper,presents,hybrid,combined,synthetic,minority,technique,random,search,RS,hyper,parameters,optimization,gradient,boosting,tree,achieve,efficient,accurate,identification,thirteen,dimensional,database,consist,basic,vector,features,established,from,image,samples,All,points,sified,either,divide,ultimate,results,into,candidate,It,found,that,technology,effectively,improve,performance,by,recommending,optimized,imbalance,ratio,Then,sixteen,classifiers,generated,four,machine,ML,models,applied,comparison,reveal,proposed,outperforms,other,fifteen,algorithms,both,classifications,Finally,discussions,importance,generalization,ability,error,conducted,experimental,indicate,more,critical,affecting,primarily,Besides,cleaning,up,sedimentary,reducing,area,fractured,contribute,improving,overall,method,provides,new,alternative,approach

0.554234