JAXA Repository / AIREX 未来へ続く、宙(そら)への英知

このアイテムに関連するファイルはありません。

タイトルApplications of Some Artificial Intelligence Methods to Satellite Soundings
著者(英)Munteanu, M. J.; Jakubowicz, O.
著者所属(英)NASA Goddard Space Flight Center
発行日1985-01-01
言語eng
内容記述Hard clustering of temperature profiles and regression temperature retrievals were used to refine the method using the probabilities of membership of each pattern vector in each of the clusters derived with discriminant analysis. In hard clustering the maximum probability is taken and the corresponding cluster as the correct cluster are considered discarding the rest of the probabilities. In fuzzy partitioned clustering these probabilities are kept and the final regression retrieval is a weighted regression retrieval of several clusters. This method was used in the clustering of brightness temperatures where the purpose was to predict tropopause height. A further refinement is the division of temperature profiles into three major regions for classification purposes. The results are summarized in the tables total r.m.s. errors are displayed. An approach based on fuzzy logic which is intimately related to artificial intelligence methods is recommended.
NASA分類METEOROLOGY AND CLIMATOLOGY
レポートNO85N29446
権利No Copyright
URIhttps://repository.exst.jaxa.jp/dspace/handle/a-is/64942


このリポジトリに保管されているアイテムは、他に指定されている場合を除き、著作権により保護されています。