FP-nets as novel deep networks inspired by vision

Philipp Grüning*, Thomas Martinetz, Erhardt Barth

*Corresponding author for this work


Feature-product networks (FP-nets) are inspired by end-stopped cortical cells with FP-units that multiply theoutputs of two filters.We enhance state-of-the-art deepnetworks, such as the ResNet and MobileNet, withFP-units and show that the resulting FP-nets performbetter on the Cifar-10 and ImageNet benchmarks.Moreover, we analyze the hyperselectivity of the FP-netmodel neurons and show that this property makesFP-nets less sensitive to adversarial attacks and JPEGartifacts.We then show that the learned model neuronsare end-stopped to different degrees and that theyprovide sparse representations with an entropy thatdecreases with hyperselectivity

Original languageEnglish
JournalJournal of Vision
Issue number1
Pages (from-to)8
Number of pages20
Publication statusPublished - 04.01.2022

Research Areas and Centers

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


Dive into the research topics of 'FP-nets as novel deep networks inspired by vision'. Together they form a unique fingerprint.

Cite this