摘要
烟丝化学成分可以由实验方法提取出来,但其关键致香成分很难确定。针对这一问题,一般采用化学分析方法,但分析耗时比较长。根据烟丝化学成分与香气风格关系,使用基于信息增益的特征选择方法进行提取致香成分。通过计算烟丝化学成分中的每个属性的信息增益,从中挑选出信息增益大于0的值作为特征选择的结果进行分类预测。实验结果表明,使用该方法能够得到较准确的关键致香成分,与传统特征选择方法相比,其特征数据集的分类结果也更加准确,可以作为烟叶香型分类的有效工具。
Chemical composition of shredded tobacco can be extracted by experimental method, but its key aroma compo- nents are difficult to determine. In view of this problem, the chemical analysis method is often adopted, but it is time-consu- ming. A feature selection method based on information gain is adopted to extract the aroma components according to the rela- tionship between chemical composition and aroma style. The information gain greater than zero as the result of feature selec- tion is picked out to conduct classification prediction by calculating the information gain of each attribute in chemical compositions. The experimental results show that the method can get more accurate key aroma components, and classification result of its feature set is more accurate than that by traditional feature selection methods. It can be used as an effective tool of tobac- co aroma classification.
出处
《现代电子技术》
2012年第18期92-94,共3页
Modern Electronics Technique
关键词
信息增益
特征选择
致香成分
烟叶香型
information gain
feature selection
aroma component
tobacco aroma
