Identification of Ancient Ceramics Based on Support Vector Machine
Jingwen Xia, Bingxiang Liu, Jiwen Zou, Zhijian Yu · 2021 International Conference on Computer Information Science and Artificial Intelligence (CISAI) · 2021
Ancient ceramics have condensed the wisdom and strength of several generations of working people in China, and have both artistic value, aesthetic value, and commercial value. However, ancient ceramics have a long history. How to accurately dating is a key problem and difficult problem. In this paper, we select ceramics from the three dynasties of Yuan, Qing and Ming as three classes, and extract the percentage of 17 materials as features. First, in order to solve the possible outliers in the features, the isolation forest is used to clearly isolate the abnormal samples of the dataset without considering the labels. Next, the classification experiment is based on the SVM model, and four kernel functions processing samples are employed, and further introduced the PSO algorithm to optimize the parameter combination of SVM, and achieved better performance.