本文關(guān)鍵詞： 選擇性集成學(xué)習(xí) 螢火蟲群優(yōu)化算法 混合核SVM 煙草香級分類　出處：《數(shù)學(xué)的實(shí)踐與認(rèn)識》2017年20期 　論文類型：期刊論文

【摘要】：針對煙草化學(xué)成分與卷煙制品香級之間確定的數(shù)學(xué)模型難以建立的問題.提出了一種基于螢火蟲群優(yōu)化算法的煙草香級集成分類方法.方法首先使用混合核SVM獨(dú)立訓(xùn)練多個(gè)個(gè)體支持向量機(jī),然后利用改進(jìn)的離散型螢火蟲群優(yōu)化算法選擇部分精度較高、差異度較大的個(gè)體分類器參與集成,最后通過多數(shù)投票法得到最終的分類預(yù)測結(jié)果.對比實(shí)驗(yàn)結(jié)果表明,算法在分類準(zhǔn)確度上具有較大的優(yōu)勢,證明了算法的有效性·從而為煙草的香級分類提供了可靠依據(jù).
[Abstract]:In order to solve the problem that it is difficult to establish a mathematical model between the chemical composition of tobacco and the aroma grade of cigarette products, an integrated classification method of tobacco aroma grade based on firefly swarm optimization algorithm is proposed. The hybrid kernel SVM is used to separate the tobacco aroma grade. Stand training multiple individual support vector machines, Then the improved discrete firefly swarm optimization algorithm is used to select individual classifiers with higher accuracy and greater difference to participate in the ensemble. Finally, the final classification prediction results are obtained by majority voting. The algorithm has a great advantage in classification accuracy, which proves the validity of the algorithm and provides a reliable basis for tobacco aroma classification.
【作者單位】： 湖南環(huán)境生物職業(yè)技術(shù)學(xué)院生態(tài)宜居學(xué)院;中南林業(yè)科技大學(xué)計(jì)算機(jī)與信息工程學(xué)院;
【基金】：湖南省自然科學(xué)基金項(xiàng)目(10JJ3066)
【分類號】：TP18;TS41
，

本文編號：1506804