1 简介

为了提高极限学习机(ELM)数据分类的精度,提出了海洋捕食者算法(SOA)的ELM分类器参数优化方法(MPA-KELM),将CV训练所得多个模型的平均精度作为MPA的适应度评价函数,为ELM的参数优化提供评价标准,用获得MPA优化最优参数的ELM算法进行数据分类.利用UCI中数据集进行仿真.

2 部分代码

%_________________________________________________________________________%  Marine Predators Algorithm source code (Developed in MATLAB R2015a)%function [Top_predator_fit,Top_predator_pos,Convergence_curve]=MPA(SearchAgents_no,Max_iter,lb,ub,dim,fobj)Top_predator_pos=zeros(1,dim);Top_predator_fit=inf; Convergence_curve=zeros(1,Max_iter);stepsize=zeros(SearchAgents_no,dim);fitness=inf(SearchAgents_no,1);Prey=initialization(SearchAgents_no,dim,ub,lb); Xmin=repmat(ones(1,dim).*lb,SearchAgents_no,1);Xmax=repmat(ones(1,dim).*ub,SearchAgents_no,1); Iter=0;FADs=0.2;P=0.5;while Iter<Max_iter         %------------------- Detecting top predator -----------------     for i=1:size(Prey,1)      Flag4ub=Prey(i,:)>ub;    Flag4lb=Prey(i,:)<lb;        Prey(i,:)=(Prey(i,:).*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;                        fitness(i,1)=fobj(Prey(i,:));     if fitness(i,1)<Top_predator_fit       Top_predator_fit=fitness(i,1);       Top_predator_pos=Prey(i,:);     end           end     %------------------- Marine Memory saving ------------------- if Iter==0   fit_old=fitness;    Prey_old=Prey; end  Inx=(fit_old<fitness);  Indx=repmat(Inx,1,dim);  Prey=Indx.*Prey_old+~Indx.*Prey;  fitness=Inx.*fit_old+~Inx.*fitness;  fit_old=fitness;    Prey_old=Prey;     %------------------------------------------------------------   Elite=repmat(Top_predator_pos,SearchAgents_no,1);  %(Eq. 10) CF=(1-Iter/Max_iter)^(2*Iter/Max_iter); RL=0.05*levy(SearchAgents_no,dim,1.5);   %Levy random number vector RB=randn(SearchAgents_no,dim);          %Brownian random number vector  for i=1:size(Prey,1)     for j=1:size(Prey,2)               R=rand();          %------------------ Phase 1 (Eq.12) -------------------       if Iter<Max_iter/3          stepsize(i,j)=RB(i,j)*(Elite(i,j)-RB(i,j)*Prey(i,j));                              Prey(i,j)=Prey(i,j)+P*R*stepsize(i,j);          %--------------- Phase 2 (Eqs. 13 & 14)----------------       elseif Iter>Max_iter/3 && Iter<2*Max_iter/3         if i>size(Prey,1)/2            stepsize(i,j)=RB(i,j)*(RB(i,j)*Elite(i,j)-Prey(i,j));            Prey(i,j)=Elite(i,j)+P*CF*stepsize(i,j);         else            stepsize(i,j)=RL(i,j)*(Elite(i,j)-RL(i,j)*Prey(i,j));                                Prey(i,j)=Prey(i,j)+P*R*stepsize(i,j);           end           %----------------- Phase 3 (Eq. 15)-------------------       else           stepsize(i,j)=RL(i,j)*(RL(i,j)*Elite(i,j)-Prey(i,j));           Prey(i,j)=Elite(i,j)+P*CF*stepsize(i,j);         end        end                                          end         %------------------ Detecting top predator ------------------          for i=1:size(Prey,1)      Flag4ub=Prey(i,:)>ub;      Flag4lb=Prey(i,:)<lb;      Prey(i,:)=(Prey(i,:).*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;    fitness(i,1)=fobj(Prey(i,:));      if fitness(i,1)<Top_predator_fit         Top_predator_fit=fitness(i,1);         Top_predator_pos=Prey(i,:);      end      end     %---------------------- Marine Memory saving ---------------- if Iter==0    fit_old=fitness;    Prey_old=Prey; end    Inx=(fit_old<fitness);    Indx=repmat(Inx,1,dim);    Prey=Indx.*Prey_old+~Indx.*Prey;    fitness=Inx.*fit_old+~Inx.*fitness;    fit_old=fitness;    Prey_old=Prey;     %---------- Eddy formation and FADs? effect (Eq 16) -----------  if rand()<FADs     U=rand(SearchAgents_no,dim)<FADs;                                                                                                   Prey=Prey+CF*((Xmin+rand(SearchAgents_no,dim).*(Xmax-Xmin)).*U);  else     r=rand();  Rs=size(Prey,1);     stepsize=(FADs*(1-r)+r)*(Prey(randperm(Rs),:)-Prey(randperm(Rs),:));     Prey=Prey+stepsize;  end  Iter=Iter+1;    Convergence_curve(Iter)=Top_predator_fit; end

3 仿真结果

4 参考文献

[1]郁智博. 基于模糊神经网络和ELM的分类算法的研究[D]. 东北大学, 2013.

博主简介：擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真，相关matlab代码问题可私信交流。

部分理论引用网络文献，若有侵权联系博主删除。