Over-MAP: Structural Attention Mechanism and Automated Semantic Segmentation Ensembled for Uncertainty Prediction

Authors

  • Charles A. Kantor KLASS, AI Research (AIR) MILA, Quebec Artificial Intelligence Institute ECP, Paris-Saclay University
  • Léonard Boussioux KLASS, AI Research (AIR) MIT ECP, Paris-Saclay University
  • Brice Rauby ECP, Paris-Saclay University Polytechnique Montréal
  • Hugues Talbot ECP, Paris-Saclay University INRIA, Paris

Keywords:

Computer Vision, Deep Learning, Fine-grained Classification, Attention Mechanisms

Abstract

Both theoretical and practical problems in deep learning classification require solutions for assessing uncertainty prediction but current state-of-the-art methods in this area are computationally expensive. In this paper, we propose a new confidence measure dubbed Over-MAP that utilizes a measure of overlap between structural attention mechanisms and segmentation methods, that is of particular interest in accurate fine-grained contexts. We show that this classification confidence increases with the degree of overlap. The associated confidence and identification tools are conceptually simple, efficient, and of high practical interest as they allow for weeding out misleading examples in training data. Our measure is currently deployed in the real-world on widely used platforms to annotate large-scale data efficiently.

Downloads

Published

2021-05-18

How to Cite

Kantor, C. A., Boussioux, L., Rauby, B., & Talbot, H. (2021). Over-MAP: Structural Attention Mechanism and Automated Semantic Segmentation Ensembled for Uncertainty Prediction. Proceedings of the AAAI Conference on Artificial Intelligence, 35(17), 15316-15322. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/17798

Issue

Section

IAAI Technical Track on Emerging Applications of AI