How to Observe Probability Distribution of Self-Information as viewed under different resolutions of $n^{-1}$ and $n^{-1/2}$ | Toshiyasu Matsushima Laboratory

Title	How to Observe Probability Distribution of Self-Information as viewed under different resolutions of $n^{-1}$ and $n^{-1/2}$ (in Japanese)
Authors	Toshiyasu Matsushima
Released Year	2014
Format	Conference
Category	Source coding
Jounal Name
Jounal Page	vol.114, no.138, IT2014-26, pp.83-88
Published Year	2014
Published Month	7
Abstract （English）	Several problems in Information Theory are investigated by asymptotic approximations of the probability of self-information or self-mutual information. Two kind of normal asymptotic approximation is discussed in this paper. One is the approximation of tail distribution, and the other is the approximation of top distribution that is the neighborhood of the mode in the distribution. Error probability and overflow probability are evaluated by the approximation of tail distribution. Mean code length and point wise code word length are evaluated by the approximation of top distribution. In this paper, a brief discussion of these approximations is presented in a cross-sectoral manner.
Note （English）	1
Manuscript
Presentation