1、,武汉大学电子信息学院,IPL,第八章 人工神经网络,模式识别与神经网络 Pattern Recognition and Neural Network,竞争学习、自组织特征映射,内容目录,IPL,第八章 人工神经网络,1. Introduction,3,2,4,5,2. Competitive Learning,3. Self-Organizing Maps,4. Learning Vector Quantization Networks,1,5. Summary,6,第八章 神经网络,3,1. Introduction,Self-organizing in networks is one o
2、f the most fascinating topics in the neural network field. Such networks can learn to detect regularities and correlations in their input and adapt their future responses to that input accordingly. The neurons of competitive networks learn to recognize groups of similar input vectors (clustering).,第
3、八章 神经网络,4,2. Competitive networks,Architecture:,The |dist| box:,第八章 神经网络,5,侧向抑制性连接,输出神经元之间有侧向抑制性连接，较强单元获胜并抑制其他单元，独处激活状态（Winner takes all, WTA）,竞争网络,第八章 神经网络,6,权值的学习规则,Kohonen Learning Rule (learnk) :,Every cluster of similar input vectors will have a neuron that outputs 1 when a vector in the cluste
4、r is presented, while outputting a 0 at all other times. The competitive network learns to categorize the input vectors.,竞争网络,第八章 神经网络,7,偏置的学习规则,Dead neurons： some neuron weight vectors may never win the competition. These unfortunate neurons never perform a useful function. Bias Learning Rule (lear
5、ncon)：biases of neurons that havent responded very frequently will increase versus biases of neurons that have responded frequently. Biases give neurons that only win the competition rarely an advantage over neurons that win often.,竞争网络,第八章 神经网络,8,竞争网络举例,Suppose we want to divide the following four
6、two-element vectors into two classes. p = .1 .8 .1 .9; .2 .9 .1 .8 p = 0.1000 0.8000 0.1000 0.9000 0.2000 0.9000 0.1000 0.8000 create a network： net = newc(0 1; 0 1,2); Training： net.trainParam.epochs = 500 net = train(net,p); During each epoch, a single vector is chosen randomly and presented to th
7、e network and weight and bias values are updated accordingly.,竞争网络,第八章 神经网络,9,竞争网络举例,Simulate the network, and finally convert its output vectors to class indices. a = sim(net,p); ac = vec2ind(a) ac = 1 2 1 2,竞争网络,Graphical example: democ1,第八章 神经网络,10,3. Self-Organizing Maps,Self-organizing feature
8、maps (SOFM) learn to classify input vectors according to how they are grouped in the input space. They differ from competitive layers in that neighboring neurons in SOM learn to recognize neighboring sections of the input space. Thus, self-organizing maps learn both the distribution and topology of the input vectors,The |dist| box:,
