Web: http://arxiv.org/abs/2102.11582

Jan. 31, 2022, 2:11 a.m. | Jishnu Mukhoti, Andreas Kirsch, Joost van Amersfoort, Philip H.S. Torr, Yarin Gal

cs.LG updates on arXiv.org arxiv.org

Reliable uncertainty from deterministic single-forward pass models is sought
after because conventional methods of uncertainty quantification are
computationally expensive. We take two complex single-forward-pass uncertainty
approaches, DUQ and SNGP, and examine whether they mainly rely on a
well-regularized feature space. Crucially, without using their more complex
methods for estimating uncertainty, a single softmax neural net with such a
feature-space, achieved via residual connections and spectral normalization,
*outperforms* DUQ and SNGP's epistemic uncertainty predictions using simple
Gaussian Discriminant Analysis *post-training* as …

arxiv deep uncertainty

More from arxiv.org / cs.LG updates on arXiv.org

Data Architect – Public Sector Health Data Architect, WWPS

@ Amazon.com | US, VA, Virtual Location - Virginia

[Job 8224] Data Engineer - Developer Senior

@ CI&T | Brazil

Software Engineer, Machine Learning, Planner/Behavior Prediction

@ Nuro, Inc. | Mountain View, California (HQ)

Lead Data Scientist

@ Inspectorio | Ho Chi Minh City, Ho Chi Minh City, Vietnam - Remote

Data Engineer

@ Craftable | Portugal - Remote

Sr. Data Scientist, Ads Marketplace Analytics

@ Reddit | Remote - United States