[Statlist] Research Webinar in Statistics *FRIDAY, 4 DECEMBER 2020* GSEM, University of Geneva

[gsem-support-instituts]

Dear All,

We are pleased to invite you to our next Research Webinar.

Looking forward to seeing you


Organizers :                                                                                   
E. Cantoni - S. Engelke - D. La Vecchia - E. Ronchetti
S. Sperlich - F. Trojani - M.-P. Victoria-Feser


FRIDAY, 4 DECEMBER 2020 at 11:15am
ONLINE
Please join the Zoom research webinar: https://unige.zoom.us/j/99238951053?pwd=dkd5UlRlYXkvNzZicnY0UlBCeW5rdz09
Password: 419459


Forecasting Time Series with Neural Networks
(joint with Nathawut Phandoidaen and Moritz Haas (both from the University of Heidelberg)
Stefan RICHTER (https://stat.math.uni-heidelberg.de/people_detail.php?id=43) - University of Heidelberg, Germany

ABSTRACT:
In this talk, I will present some theoretical contributions on forecasting high-dimensional time series with neural networks. Two forecasting methods are considered.

The first method imposes a nonlinear, encoder-decoder autoregressive structure on the time series evolution. We use a fully connected encoder-decoder network to provide a one-step forecast.

The second method uses Wasserstein generative adversarial networks (WGANs) to estimate the distribution of the conditional distribution of the next observation given the past. The behavior of the estimates is analyzed in an application with temperature data.

Under structural conditions and smoothness assumptions on the underlying evolution of the time series, we prove convergence rates which do not suffer from the curse of dimension. To do so, we use and provide new results in empirical process theory for dependent data.


Visit the website: https://www.unige.ch/gsem/en/research/seminars/rcs/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: RCS_Seminar_RICHTER.pdf
Type: application/pdf
Size: 2082680 bytes
Desc: RCS_Seminar_RICHTER.pdf
URL: <https://stat.ethz.ch/pipermail/statlist/attachments/20201130/25153bc3/attachment.pdf>