A hierarchical selective ensemble randomized neural network hybridized with heuristic feature selection for estimation of sea-ice thickness

详细信息    查看全文

• 作者：Ahmad Mozaffari ; K. Andrea Scott ; Nasser L. Azad…
• 关键词：Sea ; ice thickness ; Feature selection ; Randomized neural network ; Hierarchical selective ensemble design ; Particle swarm optimization
• 刊名：Applied Intelligence
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：46
• 期：1
• 页码：16-33
• 全文大小：
• 刊物类别：Computer Science
• 刊物主题：Artificial Intelligence (incl. Robotics); Mechanical Engineering; Manufacturing, Machines, Tools;
• 出版者：Springer US
• ISSN：1573-7497
• 卷排序：46

文摘

In this paper, a hybrid intelligent system is developed to estimate sea-ice thickness along the Labrador coast of Canada. The developed intelligent system consists of two main parts. The first part is a heuristic feature selection algorithm used for processing a database to select the most effective features. The second part is a hierarchical selective ensemble randomized neural network (HSE-RNN) that is used to create a nonlinear map between the selected features and sea-ice thickness. The required data for processing have been collected from two sensors, i.e. moderate resolution imaging spectro-radiometer (MODIS), and advanced microwave scanning radiometer-earth (AMSR-E) observing system. To evaluate the computational advantages of the proposed intelligent framework, it is given brightness temperatures data captured at two different frequencies (low frequency, 6.9GHz, and high frequency, 36.5GHz) in addition to both atmospheric and oceanic variables from forecasting models. The obtained results demonstrate the computational power of the developed intelligent algorithm for the estimation of sea-ice thickness along the Labrador coast.