【摘要】：隨著信息科技的快速發(fā)展,計(jì)算機(jī)中的經(jīng)典算法在葡萄酒產(chǎn)業(yè)中得到了廣泛的研究與應(yīng)用。機(jī)器學(xué)習(xí)算法的特點(diǎn)是運(yùn)用人工智能技術(shù),在經(jīng)過(guò)大量的樣本集訓(xùn)練和學(xué)習(xí)后可以自動(dòng)地找出運(yùn)算所需要的參數(shù)和模型。針對(duì)數(shù)據(jù)挖掘中常用的機(jī)器學(xué)習(xí)算法進(jìn)行相關(guān)的研究。以分類算法為例進(jìn)行數(shù)據(jù)挖掘技術(shù)的研究。針對(duì)SVM(支持向量機(jī))泛化能力弱的缺點(diǎn),給出了一種改進(jìn)的SVM-NSVM,即先對(duì)訓(xùn)練集進(jìn)行精選,根據(jù)每個(gè)樣本與最近鄰類標(biāo)的異同判斷樣本點(diǎn)的取舍,然后再用SVM訓(xùn)練得到分類器。針對(duì)kNN(k-最近鄰)訓(xùn)練數(shù)據(jù)集大的缺點(diǎn),給出了一種改進(jìn)的通過(guò)漸進(jìn)的思想來(lái)尋找最近鄰點(diǎn)。實(shí)驗(yàn)表明,與SVM相比,NSVM在分類正確率、分類速度上有一定的優(yōu)勢(shì)。改進(jìn)的kNN算法的復(fù)雜度明顯降低。此外,設(shè)計(jì)了葡萄酒信息數(shù)據(jù)分析系統(tǒng),利用數(shù)據(jù)挖掘方法對(duì)極大量的葡萄酒信息數(shù)據(jù)進(jìn)行分析、對(duì)比與匹配,從而可挖掘葡萄酒的主要成分對(duì)比信息和營(yíng)銷潛在信息等;再對(duì)這些成分進(jìn)行相應(yīng)的分析,并與高質(zhì)量葡萄酒中的成分進(jìn)行相應(yīng)的對(duì)比,最終得出葡萄酒的相關(guān)分析信息數(shù)據(jù),其可幫助葡萄酒生產(chǎn)廠商對(duì)葡萄酒的成分含量、品質(zhì)進(jìn)行分析。
[Abstract]:With the rapid development of information technology, the classical computer algorithms have been widely studied and applied in wine industry. The characteristic of machine learning algorithm is that the parameters and models needed for operation can be found automatically by using artificial intelligence technology after training and learning a large number of sample sets. This paper focuses on the machine learning algorithms commonly used in data mining. The classification algorithm is taken as an example to study the data mining technology. Aiming at the weakness of the generalization ability of SVM (support Vector Machine), an improved SVM-NSVM, is presented to select the training set first and judge the choice of the sample points according to the similarities and differences between each sample and the nearest neighbor. Then the classifier is trained by SVM. In view of the disadvantage of large kNN (k- nearest neighbor) training data set, an improved approach to finding nearest neighbor points by progressive thinking is presented. Experimental results show that NSVM has some advantages in classification accuracy and classification speed compared with SVM. The complexity of the improved kNN algorithm is obviously reduced. In addition, a wine information data analysis system is designed, and a large number of wine information data are analyzed, compared and matched by the method of data mining, so that the main components of wine contrast information and marketing potential information can be mined. Then these components are analyzed and compared with those in high quality wine. Finally, the relevant analysis information data of wine can be obtained, which can help the wine producers to know the composition content of wine. Quality analysis.
【作者單位】： 寧夏大學(xué)信息工程學(xué)院;
【基金】：寧夏科技支撐計(jì)劃項(xiàng)目(2015BY115) 寧夏大學(xué)研究生創(chuàng)新項(xiàng)目(GIP201625)資助
【分類號(hào)】：TP311.13

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 樊凌凌;張蕾;;基于RFID技術(shù)的葡萄酒包裝技術(shù)的研究[J];包裝工程;2008年07期

相關(guān)碩士學(xué)位論文 前1條

1 賀智文;葡萄酒俱樂(lè)部管理信息系統(tǒng)分析與設(shè)計(jì)[D];蘭州大學(xué);2013年

，

本文編號(hào)：2336285