這篇論文本基於橢圓曲線迪菲-赫爾金鑰交換協議 (ECDH)，測試類神經網路在讀取 加密訊息與過程中的公開資訊後，是否可以對原本未加密的訊息進行最低有效位 (Least Significant Bit) 的判斷，並討論對於驗證資料類神經網路模型的表現。本文將 問題歸類為深度學習中的分類問題 (Classification)，且使用深度學習中的監督式學習 (Supervised learning)。過程中為使模型最佳化順利，我們將輸入資料進行了二進位轉 換。根據實驗結果，模型可成功判斷最低有效位，且模型輸出的正確率高於 0.97。

In this paper, we study the performance of deep learning on the computation of the least significant bit derived from the elliptic curves cryptography. Here we regard the prob- lem as a classification problem and use the supervised learning in deep learning. In order to optimize the model smoothly, we turn the input data into binary representations. Ac- cording to the experimental results, the model can successfully judge the least significant bit and the accuracy is higher than 0.97.

[1] D. Coppersmith, I. Shparlinski (2000). On polynomial approximation of the discrete logarithm and the Diffie–Hellman mapping. Journal of Cryptology, vol.13, pp. 339- 360.
[2] E.C. Laskari, G.C. Meletiou, M.N. Vrahatis, E.C. Laskari, G.C. Meletiou, M.N. Vra- hatis (2005). Aitken and neville inverse interpolation methods over finite fields. Applied Numerical Analysis & Computational Mathematics, vol. 2, pp. 100-107.
[3] U. Maurer, S. Wolf (1999). The relationship between breaking the Diffie–Hellman protocol and computing discrete logarithms. SIAM Journal on Computing, vol. 28, pp. 1689-1721.
[4] G.C. Meletiou, G.L. Mullen (1992). A note on discrete logarithms in finite fields. Ap- plicable Algebra in Engineering, Communication and Computing, vol. 3, pp. 75-79.
[5] A. Winterhof (2002). Polynomial interpolation of the discrete logarithm. Designs, Codes and Cryptography, vol 25, pp. 63-72.
[6] E.C. Laskari, G.C. Meletiou,Y.C., Stamatiou,D.K. Tasoulis, M.N. Vrahatis (2007). As- sessing the effectiveness of artificial neural networks on problems related to elliptic curve cryptography. Mathematical and Computer Modelling, vol. 46, pp.174-179
[7] Y. Lecun, Y. Bengio, G. Hinton (2015). Deep learning. Nature, vol. 521, pp. 436-444.
[8] J. Schmidhuber (2015). Deep learning in neural networks: An overview. Neural Net-
works, vol. 61, 85–117
[9] 王智弘 (譯) 網路安全與密碼學概論 (2014) (原著:A. Forouzan (2007). Cryptog-
raphy and Network Security
[10] N. Koblitz (1987). Mathematics of Computation, vol. 48, pp. 203-209
[11] Silverman,J.H.(1986).Thearithmeticofellipticcurves.Graduatetextsinmathematics
[12] F. William (1969). Algebraic curves. An introduction to algebraic geometry
[13] Ruder, S. (2016). An overview of gradient descent optimization algorithms. ArXiv, abs/1609.04747.
[14] BEZOUTTHEOREM.Retrievedfromhttp://www.http://math.mit.edu/lguth/PolyMethod/lect13.pdf (May 06, 2020)