Litcius/Paper detail

Learning Deep Classifiers Consistent with Fine-Grained Novelty Detection

Jiacheng Cheng, Nuno Vasconcelos

202118 citationsDOI

Abstract

The problem of novelty detection in fine-grained visual classification (FGVC) is considered. An integrated understanding of the probabilistic and distance-based approaches to novelty detection is developed within the frame-work of convolutional neural networks (CNNs). It is shown that softmax CNN classifiers are inconsistent with novelty detection, because their learned class-conditional distributions and associated distance metrics are unidentifiable. A new regularization constraint, the class-conditional Gaussianity loss, is then proposed to eliminate this unidentifiability, and enforce Gaussian class-conditional distributions. This enables training Novelty Detection Consistent Classifiers (NDCCs) that are jointly optimal for classification and novelty detection. Empirical evaluations show that NDCCs achieve significant improvements over the state-of-the-art on both small- and large-scale FGVC datasets.

Topics & Concepts

Novelty detectionNoveltyArtificial intelligenceComputer scienceSoftmax functionPattern recognition (psychology)Probabilistic logicMachine learningConvolutional neural networkPhilosophyTheologyAnomaly Detection Techniques and ApplicationsData-Driven Disease SurveillanceDomain Adaptation and Few-Shot Learning
Learning Deep Classifiers Consistent with Fine-Grained Novelty Detection | Litcius