Title: Multiway delay embedding transform for tensors and its applications
In this presentation, I will introduce a multiway delay embedding transform (MDT) as an extension of delay embedding which is usually used for time series signals in dynamical system analysis. The MDT has several promising applications by combinatory using classical data models. I will show two examples: Tucker decomposition (CVPR2018) and auto-encoder (arXiv) for tensor/image completion task.
|Date||October 25, 2019 (Fri) 11:00 - 12:00|