Parameter-efficient Bayesian Neural Networks for Uncertainty-aware Depth Estimation

arXiv (Cornell University)

Abstract

State-of-the-art computer vision tasks, like monocular depth estimation (MDE), rely heavily on large, modern Transformer-based architectures. However, their application in safety-critical domains demands reliable predictive performance and uncertainty quantification. While Bayesian neural networks provide a conceptually simple approach to serve those requirements, they suffer from the high dimensionality of the parameter space. Parameter-efficient fine-tuning (PEFT) methods, in particular low-rank adaptations (LoRA), have emerged as a popular strategy for adapting large-scale models to down-stream tasks by performing parameter inference on lower-dimensional subspaces. In this work, we investigate the suitability of PEFT methods for subspace Bayesian inference in large-scale Transformer-based vision models. We show that, indeed, combining BitFit, DiffFit, LoRA, and CoLoRA, a novel LoRA-inspired PEFT method, with Bayesian inference enables more robust and reliable predictive performance in MDE.

Authors 5

1. Richard D. Paul

   Forschungszentrum Jülich · Helmholtz Munich

   Affiliation as printed

   Forschungszentrum Jülich , Jülich , Germany

   Helmholtz AI , Munich , Germany

2. Alessio Quercia Aachen

   Forschungszentrum Jülich · RWTH Aachen University

   Affiliation as printed

   Forschungszentrum Jülich , Jülich , Germany

   RWTH Aachen University , Aachen , Germany

3. Vincent Fortuin

   Helmholtz Munich · Technical University of Munich

   Affiliation as printed

   Helmholtz AI , Munich , Germany

   Technical University of Munich , Munich , Germany

4. Katharina Nöh
5. Hanno Scharr

   Forschungszentrum Jülich

   Affiliation as printed

   Forschungszentrum Jülich , Jülich , Germany