loading page

Streamlined and Resource-Efficient Predictive Uncertainty Estimation of Deep Ensemble Predictions via Regression
  • +3
  • Jordan F. Masakuna ,
  • D'Jeff K. Nkashama ,
  • Arian Soltani ,
  • Marc Frappier ,
  • Pierre-Martin Tardif ,
  • Froduald Kabanza
Jordan F. Masakuna
Author Profile
D'Jeff K. Nkashama
Author Profile
Arian Soltani
Author Profile
Marc Frappier
Author Profile
Pierre-Martin Tardif
Author Profile
Froduald Kabanza
Author Profile

Abstract

This paper highlights the contribution of utilizing ensemble deep  learning with auto-encoders (AEs) for out-of-distribution  data detection. The key innovation is treating ensemble UQ as a  regression problem, mapping uncertainty distribution to a single model,  reducing computational demands. This approach aligns well with the  ensemble of AEs’ uncertainty distribution, making it valuable for  resource-constrained systems and rapid decision-making in computational  intelligence.