主题：Adversarial Transformer for Multivariate Time Series Forecasting
时间：2026-08-23 15:30
地点：2教南228会议室
主持人：陈云 教授
王子栋.png王子栋，现任英国伦敦Brunel University讲席教授，欧洲科学院院士，欧洲科学与艺术院院士，IEEE Fellow，International Journal of Systems Science主编，Neurocomputing主编。多年来从事控制理论、机器学习、生物信息学等方面研究，在SCI刊物上发表国际论文六百余篇。现任或曾任十二种国际刊物的主编、副编辑或编委。曾任旅英华人自动化及计算机协会主席、东华大学CJ学者讲座教授、清华大学国家级专家。
报告摘要：In this presentation, we talk about the multivariate time series forecasting problem using adversarial transformer. A new transformer-based forecasting model, called Fusionformer, is discussed, which includes 1) the introduction of a segment-wise sequence embedding method; 2) the implementation of a fusion attention mechanism; and 3) the development of an adversarial learning method equipped with an auxiliary discriminator to facilitate the learning of data distribution. The proposed Fusionformer is applied to the early warning problem for slope failure in an open-pit mine. Finally, we conclude our main contributions and some future directions.