• English
  • Deutsch
  • Log In
    Password Login
    Research Outputs
    Fundings & Projects
    Researchers
    Institutes
    Statistics
Repository logo
Fraunhofer-Gesellschaft
  1. Home
  2. Fraunhofer-Gesellschaft
  3. Konferenzschrift
  4. Is it all a cluster game?
 
  • Details
  • Full
Options
2022
Conference Paper
Title

Is it all a cluster game?

Title Supplement
Exploring Out-of-Distribution Detection based on Clustering in the Embedding Space
Abstract
It is essential for safety-critical applications of deep neural networks to determine when new inputs are significantly different from the training distribution. In this paper, we explore this out-of-distribution (OOD) detection problem for image classification using clusters of semantically similar embeddings of the training data and exploit the differences in distance relationships to these clusters between in- and out-of-distribution data. We study the structure and separation of clusters in the embedding space and find that the supervised contrastive learning leads to well separated clusters while its self-supervised counterpart fails to do so. In our extensive analysis of different training methods, clustering strategies, distance metrics and thresholding approaches, we observe that there is no clear winner. The optimal approach depends on the model architecture and selected datasets for in- and out-of-distribution. While we could reproduce the outstanding results for contrastive training on CIFAR-10 as in-distribution data, we find standard cross-entropy paired with cosine similarity outperforms all contrastive training methods when training on CIFAR-100 instead. Cross-entropy provides competitive results as compared to expensive contrastive training methods.
Author(s)
Sinhamahapatra, Poulami  
Fraunhofer-Institut für Kognitive Systeme IKS  
Koner, Rajat
Ludwig-Maximilians-Univ. LMU
Roscher, Karsten  
Fraunhofer-Institut für Kognitive Systeme IKS  
Günnemann, Stephan
Technische Univ. München TUM
Mainwork
Workshop on Artificial Intelligence Safety, SafeAI 2022. Proceedings. Online resource  
Funder
Bayerisches Staatsministerium für Wirtschaft, Landesentwicklung und Energie StMWi  
Conference
Workshop on Artificial Intelligence Safety (SafeAI) 2022  
Conference on Artificial Intelligence (AAAI) 2022  
Open Access
DOI
10.24406/publica-fhg-417348
File(s)
Download (544.47 KB)
Rights
CC BY 4.0: Creative Commons Attribution
Language
English
Fraunhofer-Institut für Kognitive Systeme IKS  
Keyword(s)
  • out of distribution

  • OOD

  • image classification

  • contrastive learning

  • clustering

  • self-supervised learning

  • supervised learning

  • safety critical

  • Deep neural networks

  • DNN

  • artificial intelligence

  • AI

  • AI safety

  • Safe AI

  • Safe Intelligence

  • Cookie settings
  • Imprint
  • Privacy policy
  • Api
  • Contact
© 2024