报告人：李可
报告时间：2017年12月17日8:00开始
报告地点：数学学院A302
主办单位：中国矿业大学数学学院，徐州市工业与应用数学学会
报告摘要：
The spacecraft electrical signal characteristic data exist a large amount of data, high dimension features, computational complexity degree and low rate of identification problems. This report proposes the feature extraction method of deep neural networks (DNN) and transfer learning (TL) algorithm applying to classification and recognition of spacecraft experiment data. It is divided into two parts. The first one is the spacecraft electrical signal data classification, and in this part, the semi-supervised deep neural network is used for data feature extraction and classification. The second part introduces the classification of hyperspectral data, in this part transfer learning was used to solve the problem of insufficient data samples, and deep learning algorithm was used to build feature extraction classification model. The experimental results show that compared with other algorithms, the proposed method shows excellent performance in accuracy, computational efficiency, stability in dealing with spacecraft electrical signal data.
报告人简介：
北京航空航天大学人机工效与环境控制国防重点学科实验室副教授、博士，发表学术论文40余篇，获批国家发明专利五项。主持国家自然科学基金面上项目、航空科学基金、航天科技基金、凡舟基金及航空、航天院所等多项课题研究。获得国防科学技术进步一等奖一项。获得北京市德育先进工作者称号。现任中国人工智能学会认知系统与信息处理专业委员会委员。