Bio-inspired Min-Nets Improve the Performance and Robustness of Deep Networks

Abstract

Min-Nets are inspired by end-stopped cortical cells with units that output the minimum of two learned filters. We insert such Min-units into state-of-the-art deep networks, such as the popular ResNet and DenseNet, and show that the resulting Min-Nets perform better on the Cifar-10 benchmark. Moreover, we show that Min-Nets are more robust against JPEG compression artifacts. We argue that the minimum operation is the simplest way of implementing an AND operation on pairs of filters and that such AND operations introduce a bias that is appropriate given the statistics of natural images.
OriginalspracheEnglisch
TitelSVRHM 2021 Workshop NeurIPS
Erscheinungsdatum2021
PublikationsstatusVeröffentlicht - 2021

Strategische Forschungsbereiche und Zentren

  • Zentren: Zentrum für Künstliche Intelligenz Lübeck (ZKIL)
  • Querschnittsbereich: Intelligente Systeme

DFG-Fachsystematik

  • 4.43-05 Bild- und Sprachverarbeitung, Computergraphik und Visualisierung, Human Computer Interaction, Ubiquitous und Wearable Computing
  • machine learning
  • neural networks

Fingerprint

Untersuchen Sie die Forschungsthemen von „Bio-inspired Min-Nets Improve the Performance and Robustness of Deep Networks“. Zusammen bilden sie einen einzigartigen Fingerprint.

Zitieren