Questioning Ideas in Uncertainty Estimation in Deep Learning

Add to your list(s) Download to your calendar using vCal

• Guoxuan Xia (Imperial College London)
• Monday 13 November 2023, 12:00-13:00
• Hybrid: JDB Teaching Room, Engineering Department or Zoom: https://cam-ac-uk.zoom.us/meeting/register/tZ0tfuCvrDgjG9WNY52TL0qsu7PqO484Pad5.

If you have a question about this talk, please contact Simon Webster McKnight.

“Stories” in research often are simple, clean and convenient. This talk is about some of my experiences questioning some of these stories in the field of uncertainty estimation in deep learning. We will have a look at the following questions:

• Is ensemble diversity really useful for uncertainty estimation?
• Why do we ignore incorrect in-distribution predictions when evaluating Out-of-Distribution Detection?
• Are Deep Ensembles really too computationally costly?

Speaker bio: Guoxuan Xia is currently a PhD student at the Circuits and Systems group in Imperial College London. He undertook his master’s project under the supervision of Prof. Mark Gales at CUED . His research interests are primarily in the areas of reliability (uncertainty, robustness) and computational efficiency (dynamic neural networks, quantisation, knowledge distillation) in deep learning.

This talk is part of the CUED Speech Group Seminars series.

This talk is included in these lists:

• CUED Speech Group Seminars
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge talks
• Chris Davis' list
• Guy Emerson's list
• Hybrid: JDB Teaching Room, Engineering Department or Zoom: https://cam-ac-uk.zoom.us/meeting/register/tZ0tfuCvrDgjG9WNY52TL0qsu7PqO484Pad5
• Information Engineering Division seminar list
• PhD related

Note that ex-directory lists are not shown.