在許多應用中，系統需要在不會破壞舊有的資訊的前提下，能夠快速地學習新的資訊和微調舊有的資訊，這就是所謂的線上學習的特性。對於一個有效的辨識系統來說，能具備線上學習的特性是相當吸引人的。 人體免疫系統是十分地複雜的，它的許多特性與機制吸引了許多的研究者注意，近幾年來，有很多類免疫系統(AIS)的產生，這些不同的類免疫系統採用了在人體免疫系統裡一些不同的機制，來解決所要處理的問題。 在本論文中，我們提出了一個新的線上學習的類神經模糊系統，採用了人體免疫系統中的某一些特性，我們稱此類神經模糊系統為“以類免疫系統為基礎的類神經模糊系統” 。此系統在學習的過程中，能夠以漸進式的方式來建構系統，並可以應用在圖形識別與函數逼近的問題上。除了用數個人造資料集，並且也以一些真實的資料集來測試其效能，尤其特別的是，我們也將此系統應用於背光影像的補償處理。 In some applications, systems should be able to learn new classes and refine existing classes quickly and without destroying old class information. This property is referred to as on-line learning and it is a very appealing property for an efficient pattern recognition system. The immune system is a highly complicated system. Many properties of immune systems attract a great amount of attentions from compute scientists and engineers. In recently years, many artificial immune systems have been proposed. Different artificial immunes systems are inspired by different subsets of the available metaphors. In this paper, we present an on-line learning neuro-fuzzy system which was inspired by part of the mechanisms in immune systems. We name the proposed neuro-fuzzy system as the artificial immune system based neuro-fuzzy system (AISNFS). During the learning procedure, a neuro-fuzzy system can be incrementally constructed. AISNFS can be applied in pattern recognition and function approximation problems. The performance of the propose AISNFS is evaluated by not only some artificial data sets but also some real data sets. Especially, we apply the proposed AISNFS in the compensation of backlight images.
