Research On Learning Brain Effective Connectivity Network Structure Based On Firefly Algorithm

Learning brain effective connectivity(EC)networks is an important topic within the community of human brain connectome.It is of great significance for the early diagnosis and pathological study of brain diseases to accurately identify the brain EC network structure.The unsupervised data-driven learning method based on Bayesian Network(BN)has become a new research hotspot among them.However,this learning method is prone to fall into local optimum,and the accuracy of direction recognition is low.In order to overcome the shortcomings mentioned above,some scholars have combined the swarm intelligence method of global random search with Bayesian network model learning to complete the learning of brain EC network,and obtained a high quality brain EC network.The following two aspects of research work are carried out based on Firefly algorithm(FA):(1)FA,as a new swarm intelligence algorithm,may have potential application value in learning brain EC networks.Therefore,this paper proposes a new method to learn brain EC networks by FA with a reproductive mechanism(FAR-EC).The new method use the optimization of firefly population to search the optimal brain EC networks.First,a firefly individual represents a brain EC network with a few edges,which is gradually constructed through the directional movements and random movements of the firefly individual.And then,a reproductive mechanism is employed to optimize the quality of networks after a certain number of evolution iterations.Finally,the network structure represented by the individuals with the highest absolute brightness in the population is used as the learning brain EC network.Experimental results on many simulated datasets verify the effectiveness of the reproductive mechanism,and the new algorithm has obvious advantages on the whole performance compared with other algorithms.Experimental results on real datasets also show the potential practicability of the new algorithm.(2)To overcome the shortcomings of FAR-EC algorithm,such as many parameters and high time complexity,we propose a parallel searching of double firefly populations for learning brain EC network(DFA-EC).The new method uses double population optimization to learn the brain EC networks.Firstly,the initial population is divided into an elite population and a common population.Then,the brain EC networks are gradually constructed through the directional movements of the elite population and the random movements of the common population.In the process of network constructions,the population sizes of the elite population and the common population are adjusted dynamically and the information exchange between the two populations is realized by using a population migration operation.Finally,an adaptive updating mechanism based on a diversity measure is used to dynamically update the two populations after a certain number of evolution iterations.Experiments on simulated data sets show that compared with FAR-EC algorithm,the new algorithm has faster solution speed and better solution quality.Compared with several classical algorithms,DFA-EC algorithm can get better brain EC network.Experiments on real data sets show that DFA-EC algorithm has potential application value in pathological research of AD.