Weakly supervised learning and its applications in remote sensing and digital health

報告時間：2022年10月11日（星期一）下午16：00-18：00

報告地點：曲江校區（教9-313）

報告題目：Weakly supervised learning and its applications in remote sensing and digital health

報告摘要：Sensors are regarded as the “electronic sensory organ in information era”, which has changed the live style of humans. Remote Sensing and Digital Health are two typical applications of them, and these sensor systems are yielding massive data each day. How to effectively analyze the data is the key problem for the applications. In recent years, artificial intelligence has become the most promising research topic for data analysis in these two fields. However, there are still some challenges confronting the tasks: i) Label scarcity. ii) Heavy noise. iii) Data variability and heterogeneity. Here, I will present several works from me which address these issues. Different machine learning paradigms, including semi-supervised learning-based, active learning-based and Markov random field-based approaches are designed and applied. The findings from these works point out that making full use of the unlabeled data and the prior knowledge are effective ways to improve the prediction performance and reduce the annotation cost.

報告人簡介：畢海霞，西安交通大學“青年拔尖人才計劃”入選者，西安交通大學電信學部特聘研究員，博士生導師。2003年和2006於中國海洋大學分別獲得學士和碩士學位；之後於華為和愛立信從事軟件研發工作；2018年於西安交通大學獲得博士學位；2018至2021年於英國德比大學和布裏斯托大學從事博士後研究工作。主要研究領域為機器學習算法研究及其在遙感和健康醫療領域的應用。在IEEE Trans. Image Processing、IEEE Trans. Geoscience and Remote Sensing、IEEE Journal of Biomedical and Health Informatics等國際權威學術期刊和IGARSS等國際會議發表論文30餘篇。於2020年獲得了IEEE Geoscience and Remote Sensing Letters的Best Reviewer獎項(全球共5人)。