Computer Science ›› 2026, Vol. 53 ›› Issue (9): 165-172.doi: 10.11896/jsjkx.250800029

• Database & Big Data & Data Science • Previous Articles     Next Articles

Improved Conditional Gradient Algorithm Based on Double Away-step for Solving MinimumEnclosing Ball Problem

CONG Weijie1, YUE Yuanyi2, WANG Min2   

  1. 1 School of Science,Xi'an University of Posts and Telecommunications,Xi'an 710121,China
    2 School of Computer Science and Technology,Xi'an University of Posts and Telecommunications,Xi'an 710121,China
  • Received:2025-08-07 Revised:2025-12-12 Online:2026-09-15 Published:2026-09-10
  • About author:CONG Weijie,born in 1981,Ph.D, associate professor.His main research interests include optimization theory and algorithms for computational geometry and machine learning.
    YUE Yuanyi,born in 2001,postgra-duate.His main research interest is machine learning.
  • Supported by:
    National Natural Science Foundation of China(12301593) and Natural Science Basic Research Plan in Shaanxi Province(2024JC-YBQN-0048).

Abstract: The Minimum Enclosing Ball(MEB) problem is a classical optimization problem in machine learning and computational geometry,where the objective is to find a ball with the smallest radius enclosing a given set of data points.By combining the classic away-step and the pairwise away-step directions,an improved double away-step conditional gradient(DACG) algorithm is proposed for solving the MEB problem of a set of m data points in an n-dimensional space.Compared to other variants of the double-choice conditional gradient algorithm with away-step,this improved algorithm abandons the use of the traditional toward-step direction from the original conditional gradient algorithm.Instead,at each iteration,it selects the step direction by comparing the exact objective function improvement under the classic away-step and the pairwise away-step directions.The linear polynomial time complexity of the DACG algorithm for solving the (1+)-approximate solution of the MEB problem is O(mn/∈).Numerical experimental results show that the DACG algorithm clearly improves the computational efficiency of solving the MEB problem for high-dimensional large-scale datasets.In particular,compared with the classic away-step conditional gradient algorithm,the number of iterations of the DACG algorithm can be reduced by 31.8% to 46.5%,and the running time can be saved by 30.5% to 54.5%.Moreover,numerical experiments further extended the application of the DACG algorithm to support vector data description in machine learning,demonstrating its effectiveness on real-world datasets.

Key words: Conditional gradient algorithm, Classic away-step, Pairwise away-step, Minimum enclosing ball, Support vector date description

CLC Number: 

  • TP301
[1] CONG W J,AN M Y,LI C Z.An Active-set Minimum Enclosing Ball Algorithm Based on Second-order Away-step[J].Journal of Xi'an University of Posts and Telecommunications,2024,29(3):83-89.
[2] CAVALEIRO M,ALIZADEH F.A branch-and-bound method for the minimum k-enclosing ball problem[J].Operations Research Letters,2022,50(3):274-280.
[3] CONG W J,LIU H W.Research on the Algorithm for the Minimum Enclosing Ball Problem Based on Active Set Strategy[J].Computer Science,2013,40(9):234-236,253.
[4] DENG S Z,TENG D,LI X H,et al.Spherical Regularized Support Vector Description for Visual Anomaly Detection[J].Chinese Journal of Scientific Instrument,2024,45(3):315-325.
[5] WU H N,XING H J,LI G.Deep Multiple-sphere Support Vector Data Description Based on Variational Autoencoder with Mixture-of-Gaussians Prior[J].Computer Science,2024,51(6):135-143.
[6] GU X,WANG S T.Novel Domain Transfer Learning Approach Using Minimum Enclosing Ball[J].Computer Science,2013,40(7):187-192.
[7] TSANG I W,KWOK J T,CHEUNG P.Core vector machines:Fast SVM training on very large data sets [J].Journal of Machine Learning Research,2005,6(4):363-392.
[8] LEVITIN E S,POLYAK B T.Constrained minimization me-thods[J].USSR Computational Mathematics and Mathematical Physics,1966,6(5):787-823.
[9] FRANK M,WOLFE P.An algorithm for quadratic program-ming[J].Naval Research Logistics Quarterly,1956,3(1/2):95-110.
[10] WOLFE P.Convergence theory in nonlinear programming[M] //Integer and Nonlinear Programming.Amsterdam:North-Holland Publishing Company,1970:1-36.
[11] YILDIRIM E A.Two algorithms for the minimum enclosing ball problem[J].SIAM Journal on Optimization,2008,19(3):1368-1391.
[12] LACOSTE-JULIEN S,JAGGI M.On the Global Linear Convergence of Frank-Wolfe Optimization Variants[C] //Advances in Neural Information Processing Systems.Curran Associates,Inc.,2015.
[13] CHEN P H,FAN R E,LIN C J.A study on SMO-type decomposition methods for support vector machines[J].IEEE Tran-sactions on Neural Networks,2006,17(4):893-908.
[14] CONG W J,LIU H W.An SMO-type Method for Solving the MEB Problem[J].Journal of Northwest University(Natural Science Edition),2010,40(6):965-969.
[15] CONG W J,LIU H W,YE F,et al.Rank-two update algorithms for the minimum volume enclosing ellipsoid problem[J].Computational Optimization and Applications,2012,51(1):241-257.
[16] CONG W J,WANG L,SUN H.Rank-two update algorithm versus Frank-Wolfe algorithm with away steps for the weighted Euclidean one-center problem[J].Computational Optimization and Applications,2020,75(1):237-262.
[17] ÑANCULEF R,FRANDI E,SARTORI C,et al.A novel Frank-Wolfe algorithm.Analysis and applications to large-scale SVM training[J].Information Sciences,2014,285(20):66-99.
[18] TSUJI K,TANAKA K,POKUTTA S.Pairwise ConditionalGradients without Swap Steps and Sparser Kernel Herding[C] //International Conference on Machine Learning.PMLR,2022:21864-21883.
[1] CONG Wei-jie and LIU Hong-wei. Study on Algorithm of Minimum Enclosing Ball Problem Based on Active Set Strategy [J]. Computer Science, 2013, 40(9): 234-236.
[2] YING Wen-hao and WANG Shi-tong. Large Margin and Fast Learning Model Based on Difference of Similarity [J]. Computer Science, 2013, 40(8): 239-244.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!