支持向量機(jī)非線性回歸通用matlab源碼-資料下載頁(yè)
2025-06-26 19:42本頁(yè)面
【正文】 核函數(shù)參數(shù)初始化switch TKFcase 1%線性核函數(shù) K=sum(x.*y)%沒(méi)有需要定義的參數(shù)case 2%多項(xiàng)式核函數(shù) K=(sum(x.*y)+c)^pc=Para1。%c=。p=Para2。%p=2。case 3%徑向基核函數(shù) K=exp((norm(xy))^2/(2*sigma^2))sigma=Para1。%sigma=6。case 4%指數(shù)核函數(shù) K=exp(norm(xy)/(2*sigma^2))sigma=Para1。%sigma=3。case 5%Sigmoid核函數(shù) K=1/(1+exp(v*sum(x.*y)+c))v=Para1。%v=。c=Para2。%c=0。otherwise%自定義核函數(shù)，需由用戶自行在函數(shù)內(nèi)部修改，注意要同時(shí)修改好幾處！%暫時(shí)定義為 K=exp((sum((xy).^2)/(2*sigma^2)))sigma=Para1。%sigma=8。end%%%%%數(shù)據(jù)歸一化處理[X,minX,maxX]=premnmx(X)。x=2*((xminX)./(maxXminX))1。[Y,minY,maxY]=premnmx(Y)。%%%%%計(jì)算仿真測(cè)試的輸出值l=length(Alpha)。SUM=0。for i=1:lif Flag(i)0switch TKFcase 1SUM=SUM+Alpha(i)*sum(x.*X(:,i))。case 2SUM=SUM+Alpha(i)*(sum(x.*X(:,i))+c)^p。case 3SUM=SUM+Alpha(i)*exp((norm(xX(:,i)))^2/(2*sigma^2))。case 4SUM=SUM+Alpha(i)*exp(norm(xX(:,i))/(2*sigma^2))。case 5SUM=SUM+Alpha(i)*1/(1+exp(v*sum(x.*X(:,i))+c))。otherwiseSUM=SUM+Alpha(i)*exp((sum((xX(:,i)).^2)/(2*sigma^2)))。endendendy=SUM+B。%%%%%反歸一化處理y=postmnmx(y,minY,maxY)
環(huán)評(píng)公示相關(guān)推薦
鄂ICP備17016276號(hào)-1