Latent Discriminant deterministic Uncertainty

TL;DR

This paper introduces a scalable deterministic uncertainty method that learns a discriminant latent space for high-resolution semantic segmentation, achieving competitive uncertainty estimation results across multiple vision tasks.

Contribution

It proposes a novel scalable deterministic uncertainty method that relaxes Lipschitz constraints and leverages a distinction maximization layer with trainable prototypes.

Findings

Achieves competitive uncertainty estimation compared to Deep Ensembles.

Effective for high-resolution semantic segmentation and other vision tasks.

Code available for reproducibility.

Abstract

Predictive uncertainty estimation is essential for deploying Deep Neural Networks in real-world autonomous systems. However, most successful approaches are computationally intensive. In this work, we attempt to address these challenges in the context of autonomous driving perception tasks. Recently proposed Deterministic Uncertainty Methods (DUM) can only partially meet such requirements as their scalability to complex computer vision tasks is not obvious. In this work we advance a scalable and effective DUM for high-resolution semantic segmentation, that relaxes the Lipschitz constraint typically hindering practicality of such architectures. We learn a discriminant latent space by leveraging a distinction maximization layer over an arbitrarily-sized set of trainable prototypes. Our approach achieves competitive results over Deep Ensembles, the state-of-the-art for uncertainty prediction, on image classification, segmentation and monocular depth estimation tasks. Our code is available at https://github.com/ENSTA-U2IS/LDU