摘要
带树突结构的形态感知器是用格代数的方法对样本集进行分类的。树突结构目前被认为是实现逻辑运算最基本的自主计算单元,引入树突计算过程的神经元也就更接近于实际的生物神经元,具有很强的计算能力。论文根据树突过程的几何意义,通过用树突输出的方框代替覆盖算法中的“球形领域”,提出一种适于多层形态感知器学习的算法,叫做形态交叉覆盖算法,并在此基础上给出一种带树突结构的多层形态感知器模型。三个著名分类问题的实验结果很好地证明了该模型的有效性。
Morphological perceptrons endowed with dendrites use a lattice algebra approach to learn and classify a set of patterns.Dendritic structures are now viewed as the primary autonomous computational units capable of realizing logical operations.The neural that incorporates the dendritic processes is more realistic and the computational capabilities of this enriched neuron are powerful.According to the geometrical aspects of dendrite processing,we develop a new learning algorithm for multi-layer perceptron,called morphological alternative covering algorithm,by using boxes which generated by dendrites in stead of “sphere neighborhood” in the covering algorithm.Based on the algorithm,we establish a model of multi-layer morphological perceptron endowed with dendrites and apply it to three representative hard classifi- cation problems in order to exhibit its performance.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第29期30-32,74,共4页
Computer Engineering and Applications
基金
国家自然科学基金资助项目(编号:60572074)
关键词
形态神经网络
覆盖算法
树突
morphological neural network,covering algorithm,dendrite
