Missing Data Handling in Multi-Layer Perceptron
نویسندگان
چکیده
Multi layer perceptron with back propagation algorithm is popular and more used than other neural network types in various fields of investigation as a non-linear predictor. Though MLP can solve complex and non-linear problems, it cannot use missing data for training directly. We propose a training algorithm with incomplete pattern data using conventional MLP network. Focusing on the fact that BP algorithm uses the amount of the error and its sign to modify the weights, we redefined the activation function using the minimum error for incomplete pattern and modified stopping criterion is also presented. The estimation results using proposed algorithm are shown compared to the result of MLP replacing. Proposed algorithm could learn from incomplete patterns successfully and could avoid biased learning from misestimating of missing values. Key-Words: Multi Layer Perceptron, Back propagation, Missing data
منابع مشابه
Combining Missing Data Imputation and Pattern Classification in a Multi-Layer Perceptron
Multi-Layer Perceptrons (MLPs) have been successfully applied in many pattern classification tasks. However, a drawback of these learning machines is that they cannot handle input vectors that present missing data on its features. A recommended way for dealing with missing values is imputation, i.e., to fill in missing data with plausible values. This paper presents a brief review of handling m...
متن کاملMissing Value Imputation with Unsupervised Backpropagation
Many data mining and data analysis techniques operate on dense matrices or complete tables of data. Realworld data sets, however, often contain unknown values. Even many classification algorithms that are designed to operate with missing values still exhibit deteriorated accuracy. One approach to handling missing values is to fill in (impute) the missing values. In this paper, we present a tech...
متن کاملA Provably Convergent Dynamic Training Method for Multi-layer Perceptron Networks
This paper presents a new method for training multi-layer perceptron networks called DMP1 (Dynamic Multi-layer Perceptron 1). The method is based upon a divide and conquer approach which builds networks in the form of binary trees, dynamically allocating nodes and layers as needed. The individual nodes of the network are trained using a gentetic algorithm. The method is capable of handling real...
متن کاملA New Hybrid model of Multi-layer Perceptron Artificial Neural Network and Genetic Algorithms in Web Design Management Based on CMS
The size and complexity of websites have grown significantly during recent years. In line with this growth, the need to maintain most of the resources has been intensified. Content Management Systems (CMSs) are software that was presented in accordance with increased demands of users. With the advent of Content Management Systems, factors such as: domains, predesigned module’s development, grap...
متن کاملAn experimental evaluation of ZCS-DM for the prediction of urban air quality
Understanding and forecasting urban Air Quality (AQ) is not only a multifaceted and computationally challenging problem for machine learning algorithms, but also a difficult task for human-decision makers: the strict regulatory framework, in combination with the public demand for better information services poses the need for robust, efficient and, more importantly, understandable forecasting m...
متن کامل