The lunch will be served 12:00, while the talk starts at 12:15.

Speakers: Fredrik Lundvall Wollbraaten (UiO, Doctoral Research Fellow, https://www.mn.uio.no/math/english/people/aca/fredrlwo/)

Location: UiO (8th floor of Niels Henrik Abels hus) and Zoom.

Title: Approximating multivariate point process intensities by gradient boosting

Abstract: We present a coherent framework for learning the M intensity functions describing the behavior of M jointly observed event types, by encoding the history into covariates and applying existing gradient boosting software to do the estimation. Compared to existing state-of-the-art methods using Recurrent Neural Networks, however, the method is not able to clearly beat their performance in terms of test-set log-likelihood. We suspect this to be related to the difficulty of encoding the history of the point process as a smooth and low-dimensional covariate vector; the RNN method learns this mapping automatically. While this negative result may be valuable on its own, we intend to try to use the output of the RNN method as covariates in the gradient boosting framework.

Join Zoom Meeting:
https://uio.zoom.us/j/63199595088?pwd=Y05GYkJwR1dCUHFhUnlGMFpsdGc3UT09

Meeting ID: 631 9959 5088
Passcode: 302551

Welcome!
Best regards,
Thea Roksvåg and Lars Henry Berge Olsen.

Fredrik Lundvall Wollbraaten