發明
中華民國
101138827
I 599968
找出神經網路中神經元間最短路徑之方法
國立清華大學
2017/09/21
本發明係揭露一種找出(三維或更高維度)神經網路中神經元間最短路徑之方法,其步驟包含:首先,透過處理裝置建立一三維或更高維神經空間資料庫於一儲存空間內,其中三維神經空間座標資料庫中包含複數個神經元分佈於其中;接著,透過處理裝置判定三維或更高維神經空間資料庫中之每一神經元是否與其他神經元相連結;之後,利用所有節點間最短路徑(All-pairs Shortest Paths)演算法計算出全部神經元彼此相連結之最短路徑表格並儲存於儲存空間內。如此,只要選取此三維或更高維神經空間資料庫中任二個可相連之神經元,則可透過搜尋上述最短路徑表格以得出此二個神經元間之最短路徑。 The present invention demonstrates a method to find shortest pathway between neurons in a 3D (or higher dimension) network. The steps of the method comprises: firstly, establishing a 3D or higher dimension neuron image database in a space with coordinates, which includes a plurality of neurons distributing therein, in a storage apparatus by a processing device; subsequently, determining whether each neuron in the 3D or higher dimension neuron image database has connections with other neurons by the processing device; finally, utilizing all-pairs shortest paths algorithm to construct a shortest paths table of all neurons and store the table in the storage apparatus. Thus, when any two neurons of the 3D or higher dimension neuron image database are selected, the possible shortest paths of the two neurons could be obtained by searching the shortest paths table.
智財技轉組
03-5715131-62219
版權所有 © 國家科學及技術委員會 National Science and Technology Council All Rights Reserved.
建議使用IE 11或以上版本瀏覽器,最佳瀏覽解析度為1024x768以上|政府網站資料開放宣告
主辦單位:國家科學及技術委員會 執行單位:台灣經濟研究院 網站維護:台灣經濟研究院