| Co-Investigator(Kenkyū-buntansha) |
NAGAOKA Hiroshi Univ.of Elect.-Comm., Grad.Shool of Inform.Sys., Associate Professor, 情報システム学研究科, 助教授 (80192235)
HOSHI Mamoru Univ.of Elect.-Comm., Grad.Shool of Inform.Sys., Professor, 情報システム学研究科, 教授 (80125955)
|
|---|
| Research Abstract |
(1). We have proved the coding theorem and the strong converse theorem for a very general class of communication channels without making any 'traditional' assumption such as memorylessness, stationality, etc. The proof is based on the newly introduced concepts : information spectrum and limsup (liminf) in probability.(IEEE Trans. Inform. Theory, 40, pp.1147-1157.)
(2). Suppose that a pair of correlated sources (X,Y) are modeled by a joint probability distribution p(X,Y ; rheta) including the unknown parameter rheta. In the situation where the pair of data sequences from the sources are to be transmitted subject to respective rate restrictions, what accuracy can we attain in estimating rheta by optimizing both coding scheme and estimation? This is the multiterminal estimation problem, and we have solved it on some mathematical assumptions. The research is an attempt to relate and fuse concepts in two different fields 'information theory' and 'statistics' such as Shannon information and Fisher information, universal coding and maximum-likelihood estimator, etc.(IEEE Trans. Inform. Theory, 41, pp.1802-1833.)
(3). We have also studied many subjects concerned with the present project in the fields of information theory, stochastic network models, quantum information theory, etc., which are presented in scientific journals and international conferences as shown below.
|
