本論文是針對次微米MOSFET元件的臨限電壓（threshold voltage）與幾何尺寸複雜關係，提出使用模糊理論（fuzzy theory）結合類神經網路（neural network）學習的一個新想法，用來預測金氧半場效電晶體臨限電壓對通道長度及對通道寬度的關係 。 此方法為新創意，也可以說只要存在有不確定性，使用模糊（fuzzy）理論加上神經網路（neural network）的學習方式，將可以達到良好的預測效果。

　　　　　This paper proposes a new method to predict the anomalous threshold voltage behavior in submicrometer MOSFETs by using fuzzy theory and neural network. It has been developed to analyze the threshold voltage behaviours due to the geometric effect in MOSFETs. The predict results are compared with experimental data obtained from actual devices. A good agreement has been obtained on the threshold voltage behaviors versus device geometry.