Jun glaze

In our previous work [63], PCA, a classical chemometric method, was used to find some mathematical model for the identification of ancient Jun wares. At this time, 19 ancient Jun glazes, 7 glazes of imitation products, and 18 copper red glazes were investigated. The compositions of these examples are listed in Table 11.1. 

Table 11.1 Chemical composition of Jun glazes and other glazes(wt%).

Class 1 denotes Jun glaze, class 2 denotes glaze of imitation product and class 3 denotes copper red glaze. 

Yi>1.5 the area is the distribution region of ancient Jun glaze samples. 

Fig. 11.1 classification of Jun glaze, glaze of imitation products and copper... 

Jun glaze, 2. glaze of imitation products, 3. copper red glaze. 

Since the data listed in Table 11.1 include some Jun glaze samples produced in Yuan Dynasty. We rebuild the data file shown in Table 11.2. 

Table 11.2 The composition of Jun glaze of Song dynasty and the glaze of imitation...

Using the feature selection algorithm based on SVM, it has been found that a data set having three features, MgO, T1O2 and P2O5 contents, can get 100% rate of correct prediction in LOO cross-validation test. And the criterion for ancient Jun glaze is as follows ... 

Another method of identification of Jun wares can be obtained by data processing with SVR. It can be found that the relationship between Ti02 content and the contents of the other components in ancient Jun glaze samples can be expressed by the following equation ... 

Fig. 11.2 Relationship between Ti02 content and other components in Jun glazes produced in Song dynasty.

Figure 4.21. Glazing module. [Adapted from Kobiehel, M. D. Vanarak, V. C., US Patent 5,915,780, Jun. 29, 1999.]...

---