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.
Original languageEnglish
Title of host publicationSVRHM 2021 Workshop NeurIPS
Publication date2021
Publication statusPublished - 2021

Research Areas and Centers

  • Centers: Center for Artificial Intelligence Luebeck (ZKIL)
  • Research Area: Intelligent Systems

DFG Research Classification Scheme

  • 409-05 Interactive and Intelligent Systems, Image and Language Processing, Computer Graphics and Visualisation

Fingerprint

Dive into the research topics of 'Bio-inspired Min-Nets Improve the Performance and Robustness of Deep Networks'. Together they form a unique fingerprint.

Cite this