報告題目：Principle Component Analysis and Euclidean Distance Matrix Optimization
報告摘要：Principle Component Analysis (PCA) is probably the most widely used statistical method for data analysis and dimensionality reduction. As early as in 1960s, it was claimed that Principle Co-ordinate Analysis (PCoA) is more powerful than the traditional PCA. This talks ponders on its implications and begins with a brief (technical) introduction of both PCA and PCoA, in particular, on their link to optimization. When data was grossly corrupted or missing, the robust PCA has become a major approach to recovering the true data and is surprisingly successful under reasonable conditions. In contrast, the line along PCoA has been lacking in good progress. We argue that Euclidean Distance Matrix optimization may provide a key to further developments.