Deep shared representation learning for weather elements forecasting
This paper introduces a data-driven predictive model based on deep convolutional neural networks (CNN) architecture for wind speed prediction in weather data. The model exploits the spatio-temporal multivariate weather data for learning shared representations and forecasting weather elements for a number of user defined weather stations simultaneously in an end-to-end fashion. The experiment concerns wind speed prediction at three weather stations located in Denmark for 6 and 12 hours ahead based on a high temporal resolution dataset.
Cross-Domain Neural-Kernel Networks
A novel cross-domain neural-kernel networks architecture for domain adaption problem is introduced. The proposed model consists of two stream networks corresponding to the source and target domains which are enriched with a coupling term. The introduced coupling term aims at enforcing correlations among the output of the intermediate layers of the two stream networks. Experimental results are given to illustrate the effectiveness of the proposed approaches on real-life datasets.
Siamak Mehrkanoon is currently an Assistant Professor at the Department of Data Science and Knowledge Engineering (DKE), Maastricht University, the Netherlands. He received his PhD degree in Machine Learning in 2015 at KU Leuven, Belgium. He was a visiting researcher at the Department of Automation at Tsinghua University, China, in 2014, a Postdoc Research Fellow at University of Waterloo, Canada, 2015-2016, and a Visiting Postdoc Researcher at Cognitive Systems Laboratory, University of Tubingen, Germany, in 2016 and an FWO Postdoc Research Fellow at STADIUS, KU Leuven, Belgium. His current research interests encompass deep learning, neural networks, kernel models.