本文主要在設計一自組式概率神經網路，用以有效估算連續變數冬概率。此一神經網路不僅可以根據數據來自行調整架構，並且只需固定的儲存空間。對於大量資訊之有效處理，應有很大的助益。 In this study, we first discuss the problems of classifications versus explicit probability distributions. A self-organizing probabilistic neural network with a fixed size storage is then contrived for continuous variables to effectively estimate the probability density functions The simulation results are discussed.
