Jul 2026
Rethinking Likelihood distributions: Student's t Likelihood Boosts Bayesian Neural Network Performance
It is demonstrated that Student's t yields better predictive performance than a Gaussian likelihood distribution, independent of the data distribution and MLP architecture (depth and width), and can also lead to shorter training times, while still being easy to implement.
Pei-Hsuan Hsia, L. Heyen, Arvid Weyrauch et al.
· arXiv.org · 0 citations