Abstract: Aiming at the shortcomings of slow convergence speed and easy to fall into local optimum in the optimization process of the Salp swarm algorithm (SSA),a Levy Flight-based Conditional Updating Salp Swarm Algorithm (abbreviated as LECUSSA) is proposed and it is used in the feature subset selection of classification algorithm.Firstly,the leader position is updated randomly by using the long and short jump characteristics of Levy Flight strategy,which enhances the global optimal search ability.Secondly,the conditional updating condition to the follower’s position is added to make the follower no longer follow blindly,thus accelerating the convergence speed.The performance of LECUSSA algorithm is compared with other algorithms on 23 benchmark functions.The algorithm is applied to the selection of classification feature subset of SVM algorithm,and 8 UCI datasets are used to compare the performance of the classification results after feature selection.The experimental results show that LECUSSA has good global optimal search ability and fast convergence speed.After feature selection using LECUSSA algorithm,the feature subset with the best classification accuracy can be found.

Key words: Conditional update, Feature selection, Levy flight, Salp swarm algorithm