Hyperparameter-Free Out-of-Distribution Detection Using Cosine Similarity

Engkarat Techapanurak, Masanori Suganuma, Takayuki Okatani

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The ability to detect out-of-distribution (OOD) samples is vital to secure the reliability of deep neural networks in real-world applications. Considering the nature of OOD samples, detection methods should not have hyperparameters that need to be tuned depending on incoming OOD samples. However, most recently proposed methods do not meet this requirement, leading to a compromised performance in real-world applications. In this paper, we propose a simple and computationally efficient, hyperparameter-free method that uses cosine similarity. Although recent studies show its effectiveness for metric learning, it remains uncertain if cosine similarity works well also for OOD detection. There are several differences in the design of output layers from the metric learning methods; they are essential to achieve the best performance. We show through experiments that our method outperforms the existing methods on the evaluation test recently proposed by Shafaei et al. which takes the above issue of hyperparameter dependency into account; it achieves at least comparable performance to the state-of-the-art on the conventional test, where other methods but ours are allowed to use explicit OOD samples for determining hyperparameters. Lastly, we provide a brief discussion of why cosine similarity works so well, referring to an explanation by Hsu et al.

Original languageEnglish
Title of host publicationComputer Vision – ACCV 2020 - 15th Asian Conference on Computer Vision, 2020, Revised Selected Papers
EditorsHiroshi Ishikawa, Cheng-Lin Liu, Tomas Pajdla, Jianbo Shi
PublisherSpringer Science and Business Media Deutschland GmbH
Pages53-69
Number of pages17
ISBN (Print)9783030695378
DOIs
Publication statusPublished - 2021
Event15th Asian Conference on Computer Vision, ACCV 2020 - Virtual, Online
Duration: 2020 Nov 302020 Dec 4

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12625 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th Asian Conference on Computer Vision, ACCV 2020
CityVirtual, Online
Period20/11/3020/12/4

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Hyperparameter-Free Out-of-Distribution Detection Using Cosine Similarity'. Together they form a unique fingerprint.

Cite this