RBF(徑向基神經(jīng)網(wǎng)絡(luò)）

GRNN廣義神經(jīng)網(wǎng)絡(luò)

PNN概率神經(jīng)網(wǎng)絡(luò)

RBF算例實(shí)現(xiàn)（回歸）

數(shù)據(jù)集

% 1. 導(dǎo)入數(shù)據(jù)
load spectra_data.mat
NIR輸入，octane輸出

%% I. 清空環(huán)境變量 clear all clc%% II. 訓(xùn)練集/測(cè)試集產(chǎn)生 %% % 1. 導(dǎo)入數(shù)據(jù) load spectra_data.mat%% % 2. 隨機(jī)產(chǎn)生訓(xùn)練集和測(cè)試集 temp = randperm(size(NIR,1)); % 訓(xùn)練集——50個(gè)樣本 P_train = NIR(temp(1:50),:)'; T_train = octane(temp(1:50),:)'; % 測(cè)試集——10個(gè)樣本 P_test = NIR(temp(51:end),:)'; T_test = octane(temp(51:end),:)'; N = size(P_test,2);%% III. RBF神經(jīng)網(wǎng)絡(luò)創(chuàng)建及仿真測(cè)試 %% % 1. 創(chuàng)建網(wǎng)絡(luò) net = newrbe(P_train,T_train,30);%30為speend%% % 2. 仿真測(cè)試 T_sim = sim(net,P_test);%% IV. 性能評(píng)價(jià) %% % 1. 相對(duì)誤差error error = abs(T_sim - T_test)./T_test;%% % 2. 決定系數(shù)R^2 R2 = (N * sum(T_sim .* T_test) - sum(T_sim) * sum(T_test))^2 / ((N * sum((T_sim).^2) - (sum(T_sim))^2) * (N * sum((T_test).^2) - (sum(T_test))^2)); %% % 3. 結(jié)果對(duì)比 result = [T_test' T_sim' error']%% V. 繪圖 figure plot(1:N,T_test,'b:*',1:N,T_sim,'r-o') legend('真實(shí)值','預(yù)測(cè)值') xlabel('預(yù)測(cè)樣本') ylabel('辛烷值') string = {'測(cè)試集辛烷值含量預(yù)測(cè)結(jié)果對(duì)比';['R^2=' num2str(R2)]}; title(string)

GRNN和PNN神經(jīng)網(wǎng)絡(luò)(分類）

數(shù)據(jù)集：經(jīng)典案例鳶尾花分類

feature有4維數(shù)據(jù)：花萼長(zhǎng)度，花萼寬度，花瓣長(zhǎng)度，花瓣寬度

classes花的種類
‘setosa’, ‘versicolor’, ‘virginica’

%% I. 清空環(huán)境變量 clear all clc%% II. 訓(xùn)練集/測(cè)試集產(chǎn)生 %% % 1. 導(dǎo)入數(shù)據(jù) load iris_data.mat%% % 2 隨機(jī)產(chǎn)生訓(xùn)練集和測(cè)試集 P_train = []; T_train = []; P_test = []; T_test = []; for i = 1:3temp_input = features((i-1)*50+1:i*50,:);temp_output = classes((i-1)*50+1:i*50,:);n = randperm(50);% 訓(xùn)練集——120個(gè)樣本P_train = [P_train temp_input(n(1:40),:)'];T_train = [T_train temp_output(n(1:40),:)'];% 測(cè)試集——30個(gè)樣本P_test = [P_test temp_input(n(41:50),:)'];T_test = [T_test temp_output(n(41:50),:)']; end%% III. 模型建立 result_grnn = []; result_pnn = []; time_grnn = []; time_pnn = []; for i = 1:4for j = i:4p_train = P_train(i:j,:);p_test = P_test(i:j,:);%% % 1. GRNN創(chuàng)建及仿真測(cè)試t = cputime;% 創(chuàng)建網(wǎng)絡(luò)net_grnn = newgrnn(p_train,T_train);% 仿真測(cè)試t_sim_grnn = sim(net_grnn,p_test);T_sim_grnn = round(t_sim_grnn);%取整操作，因?yàn)槭欠诸?#xff0c;目標(biāo)是整數(shù)t = cputime - t;time_grnn = [time_grnn t];result_grnn = [result_grnn T_sim_grnn'];%%% 2. PNN創(chuàng)建及仿真測(cè)試t = cputime;Tc_train = ind2vec(T_train);% 創(chuàng)建網(wǎng)絡(luò)net_pnn = newpnn(p_train,Tc_train);% 仿真測(cè)試Tc_test = ind2vec(T_test);t_sim_pnn = sim(net_pnn,p_test);T_sim_pnn = vec2ind(t_sim_pnn);t = cputime - t;time_pnn = [time_pnn t];result_pnn = [result_pnn T_sim_pnn'];end end%% IV. 性能評(píng)價(jià) %% % 1. 正確率accuracy accuracy_grnn = []; accuracy_pnn = []; time = []; for i = 1:10accuracy_1 = length(find(result_grnn(:,i) == T_test'))/length(T_test);accuracy_2 = length(find(result_pnn(:,i) == T_test'))/length(T_test);accuracy_grnn = [accuracy_grnn accuracy_1];accuracy_pnn = [accuracy_pnn accuracy_2]; end%% % 2. 結(jié)果對(duì)比 result = [T_test' result_grnn result_pnn] accuracy = [accuracy_grnn;accuracy_pnn] time = [time_grnn;time_pnn]%% V. 繪圖 figure(1) plot(1:30,T_test,'bo',1:30,result_grnn(:,4),'r-*',1:30,result_pnn(:,4),'k:^') grid on xlabel('測(cè)試集樣本編號(hào)') ylabel('測(cè)試集樣本類別') string = {'測(cè)試集預(yù)測(cè)結(jié)果對(duì)比(GRNN vs PNN)';['正確率:' num2str(accuracy_grnn(4)*100) '%(GRNN) vs ' num2str(accuracy_pnn(4)*100) '%(PNN)']}; title(string) legend('真實(shí)值','GRNN預(yù)測(cè)值','PNN預(yù)測(cè)值') figure(2) plot(1:10,accuracy(1,:),'r-*',1:10,accuracy(2,:),'b:o') grid on xlabel('模型編號(hào)') ylabel('測(cè)試集正確率') title('10個(gè)模型的測(cè)試集正確率對(duì)比(GRNN vs PNN)') legend('GRNN','PNN') figure(3) plot(1:10,time(1,:),'r-*',1:10,time(2,:),'b:o') grid on xlabel('模型編號(hào)') ylabel('運(yùn)行時(shí)間(s)') title('10個(gè)模型的運(yùn)行時(shí)間對(duì)比(GRNN vs PNN)') legend('GRNN','PNN')

代碼解析：
因?yàn)榛ㄓ腥N，所以需要分段劃分

for i = 1:3temp_input = features((i-1)*50+1:i*50,:);temp_output = classes((i-1)*50+1:i*50,:);n = randperm(50);% 訓(xùn)練集——120個(gè)樣本P_train = [P_train temp_input(n(1:40),:)'];T_train = [T_train temp_output(n(1:40),:)'];% 測(cè)試集——30個(gè)樣本P_test = [P_test temp_input(n(41:50),:)'];T_test = [T_test temp_output(n(41:50),:)']; end

因?yàn)榛ㄓ兴膫€(gè)特征，我們將兩兩特征組合

for i = 1:4for j = i:4p_train = P_train(i:j,:);p_test = P_test(i:j,:);

代碼和數(shù)據(jù)百度云

鏈接：https://pan.baidu.com/s/1Z1txLCFJ_Iif_skcb57PiQ 提取碼：yn6c 復(fù)制這段內(nèi)容后打開百度網(wǎng)盤手機(jī)App，操作更方便哦

作者：電氣工程計(jì)算機(jī)萌新：余登武

總結(jié)

以上是生活随笔為你收集整理的MATLAB机器学习系列-5 RBF、GRNN和PNN神经网络原理及其例子代码的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。