t-SNE LDA算法在仿生嗅觉中的应用研究*

将流形学习中的t-SNE算法引入仿生嗅觉领域中，本研究提出一种基于t-SNE与LDA算法相结合的气味分类鉴定新方法。由PEN3电子鼻获取物质气味特征信息，通过t-SNE算法将非线性、高维度的气味响应数据降维到低维空间，并利用LDA算法对低维数据进行分类和识别。本研究利用五种不同成分的纺织品材料气味信息，通过t-SNE、PCA LDA和t-SNE LDA三种算法作对比实验。实验结果表明，相较于其他两种方法，t-SNE LDA算法对常见的不同成分纺织品材料拥有更好的分类和识别效果，而且t-SNE LDA算法得到的结果具有较小的类内离散度和较大的类间离散度。因此，t-SNE LDA算法是仿生嗅觉中气味分类和识别的一个新方法。

Research on Bionic Olfactory Application Using t-SNE LDA

This research introduces the t-SNE, a manifold learning algorithm, into the field of bionic olfaction, and proposes a method of odor classifications and identifications using t-SNE LDA technique. Odorant characteristic data were collected from PEN3 electronic nose. Then these nonlinear and high-dimensional odorant response data were reduced to low-dimensional space by using t-SNE. Finally, LDA was used for classifications and recognitions in the low dimensional data. This research implemented comparative experiments (t-SNE, PCA LDA and t-SNE LDA) of the classifications and identification for five different ingredient textile material. The result shows that t-SNE LDA has higher ability of distinguishing and recognizing common textile ingredient materials and the output according to t-SNE LDA has smaller within-class scatter and larger between-class scatter. Consequently, t-SNE LDA is supposed to be a new method for odor classifications and identifications in bionic olfaction.