頁籤選單縮合
題 名 | 以類神經網路預測軟體可靠度=Prediction of Software Reliability Using Neural Network |
---|---|
作 者 | 蔡智勇; 薛義誠; | 書刊名 | 明志學報 |
卷 期 | 36:1 2004.06[民93.06] |
頁 次 | 頁1-10 |
分類號 | 312.49 |
關鍵詞 | 倒傳遞式類神經網路; 軟體可靠度; 軟體度量; Software reliability; Software metric; Back-propagation neural network; |
語 文 | 英文(English) |
中文摘要 | 軟體可靠度模式是用來評估系統在運作階段可能發生故障頻率的程度。並且所測量與預測結果,可以協助軟體開發人員判斷計畫的現況以及計畫的時程等管理階層上的問題。目前現存的軟體可靠度模式,都是侷限在本身的關鍵性假設。類神經網路本身具有分析模組的優點,所以它可以調整模組複雜度去配合複雜的失效歷史紀錄,因此,只要將失效歷史記錄輸入,類神經網路模組就可以自動產生內部失效處理模組及預測未來失效模組,而且不需要任何假設前題。本文將定義軟體度量中所有的輸入的變數,並且建立軟體度量可靠度模式。然後輸入軟體測試紀錄,經由模擬類神經網路的套裝軟體,建立倒傳遞式類神經網路模式來預測失效量並且建立軟體可靠度預測系統。 |
英文摘要 | The Software Reliability Modeling is to evaluate the system failure rate in operating phase. The results can help the developers to judge the status and the schedule of the plan. Since all the existing software reliability models are based on some key assumptions. Neural network model has an advantage over analysis models, so it adjusts model complexity to match the complexity of the failure history. Therefore, using failure history as input, no assumption, the neural network model automatically develops its own internal model of the failure and predicts further failures. In this research will define all of variables with software metric and to establish modular software metric reliability models. Then to load test records over neural network package software to establish back-propagation neural network models models to predict failures and establish software reliability prediction systems. |
本系統中英文摘要資訊取自各篇刊載內容。