Procedure Neural Networks: A Novel NNs Model Dealing with Time Series Problem Jonsen Liang Abstract: A novel neural network model, Procedure Neural Network (PNN), is proposed dealing with time series problems. Firstly, a new type of neuron is constructed. Secondly, several forms of procedure neural network models are presented. Thirdly, some theoretical issues are mentioned. Fourthly, supervised learning algorithm is discussed. Lastly, some time series application examples are listed. (Jonsen is a senior researcher from College of Information Science & Engineering, Zhejiang Normal University, China. He will stay at the Center for Machine Perception for one year, trying to link his expertise in neural and fuzzy computation with computer vision.)