我們提出了一種資料視覺化方法st-SNE。這個方法是建立在Van der Maaten and Hinton (2008) 的t-SNE 主軸之上，並且引入了加權分類訊息。這個重要的差異使得st-SNE 大致上保留了資料點之間的主要結構，也同時能夠在同異類區塊上做出更明確的區分。在一些標竿資料以及模擬資料上，我們進行了比較與測試。這些實驗結果顯示，st-SNE 是一個優良，易於調整且具相當後續發展潛力的視覺化方法。最後基於t-SNE 在R中的package Rtsne，我們將本研究的主要計算模擬與st-SNE 方法，整理寫成R package stsne 並置於Github 自由下載。

We propose a data visualization method st-SNE by incorporating the weighted class/label information into t-SNE of Van der Maaten and Hinton (2008). The proposed st-SNE preserves some of the main structure among the data points while delivering better classification visualization. The performances and comparisons are made based on some benchmark data sets as well as some simulated data sets. The implementation of st-SNE is packaged as a R package stsne freely available at Github.

1 緒論 3
1.1 動機 3
1.2 章節介紹 6
2 t-SNE 7
2.1 目的 7
2.2 想法 7
2.3 作法 10
3 Supervised t-SNE 13
3.1 st-SNE 13
3.2 基於距離的t-SNE 的監督式化 16
4 實驗 19
4.1 資料 19
4.2 參數設定 20
4.3 比較 22
4.4 權重調整 32
4.5 混淆矩陣 34
5 R package 37
6 結論 43
參考文獻 45
A t-SNE 梯度的推導修正 47

M. Belkin and P. Niyogi. Laplacian eigenmaps and spectral techniques for embedding and clustering. In NIPS, volume 14, pages 585–591, 2001.

P. Demartines and J. Hérault. Curvilinear component analysis: A self-organizing neural network for nonlinear mapping of data sets. IEEE Transactions on neural networks, 8(1): 148–154, 1997.

I. Goodfellow, Y. Bengio, and A. Courville. Deep Learning. MIT Press, 2016. http://www.deeplearningbook.org.

L. Hajderanj, I. Weheliye, and D. Chen. A new supervised t-SNE with dissimilarity measure for effective data visualization and classification. In Proceedings of the 2019 8th International Conference on Software and Information Engineering, pages 232–236, 2019.

B. S. Hettich S. UCI KDD archive., 1999. URL kdd.ics.uci.edu.

G. Hinton and S. T. Roweis. Stochastic neighbor embedding. In NIPS, volume 15, pages 833–840. Citeseer, 2002.

Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11):2278–2324, 1998.

S. A. Nene, S. K. Nayar, and H. Murase. Columbia object image library. 1996.

S. T. Roweis and L. K. Saul. Nonlinear dimensionality reduction by locally linear embedding. Science, 290(5500):2323–2326, 2000.

J. W. Sammon. A nonlinear mapping for data structure analysis. IEEE Transactions on computers, 100(5):401–409, 1969.

J. B. Tenenbaum, V. De Silva, and J. C. Langford. A global geometric framework for nonlinear dimensionality reduction. Science, 290(5500):2319–2323, 2000.

L. Van der Maaten. Accelerating t-SNE using tree-based algorithms. The Journal of Machine Learning Research, 15(1):3221–3245, 2014.

L. Van der Maaten and G. Hinton. Visualizing data using t-SNE. Journal of machine learning research, 9(11), 2008.

L. Van der Maaten Web. t-SNE –laurens van der maaten. URL https://lvdmaaten.github.io/tsne/.

K. Q. Weinberger, F. Sha, and L. K. Saul. Learning a kernel matrix for nonlinear dimensionality reduction. In Proceedings of the twenty-first international conference on Machine learning, page 106, 2004.

H. Wickham. R packages: organize, test, document, and share your code. O’Reilly Media, 2015. https://r-pkgs.org/.

Y.-F. Zhang and H.-M. Wu. Supervised t-SNE for dimension reduction and visualization based on class information. Journal of the Chinese Statistical Association, 59:53–97, 2021.