Research And Application Of Extreme Learning Machine Algorithm Based On Imbalance And Online Learning

As we all know, the study of meteorological satellite cloud is the main method of obtaining weather forecast and detecting global climate change. Among them, the classification of cloud map and the calculation of cloud cover have important influence on weather forecast. Therefore, the classification of meteorological satellite cloud image classification has been the focus of domestic and foreign scholars. The main purpose of satellite cloud image classification is to extract the main features of different meteorological satellite images, and good feature extraction has a great influence on the accuracy of late cloud cover calculation. However, the extraction rate of satellite cloud features is relatively low at home and abroad, which leads to the great deviation of the later cloud cover calculation. This paper, after reference to the domestic and foreign scholars related research, put forward a imbalance and online learning algorithm for Extreme Learning Machine, and used in satellite cloud image detection classification.Since 2006, the field of machine learning has made breakthrough progress;especially the Extreme Learning Machine has been widely used in many fields.Compared with the traditional algorithm model, the Extreme Learning Machine has the advantages of fast training speed and good generalization performance. The traditional neural network method does not classify the imbalance data in the satellite cloud. Therefore, this paper uses a modified extreme learning algorithm to detect and classify the imbalance samples of satellite images. The main work of this paper includes the following two parts:1. Extract samples of satellite images as training samples for extreme learning machines. Using the improved imbalance and online Extreme Learning Machine to detecting the thin cloud, thick cloud, thick cloud boundary and clear sky in the satellite cloud. The cloud diagram is compared with the traditional detection method and the advantages and disadvantages are analyzed.2. Extract new imbalanced data samples, using imbalance and online Extreme Learning Machine for the classification of cloud clouds of the satellite cloud(including cloud texture, color and shape). Compare the results of the classification with several other machine learning methods to assess its accuracy and reliability.