English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Computational models trained on a large amount of natural images are thestate-of-the-art to study human vision - usually adult vision. Computationalmodels of infant vision and its further development are gaining more and moreattention in the community. In this work we aim at the very beginning of ourvisual experience - pre- and post-natal retinal waves which suggest to be apre-training mechanism for the primate visual system at a very early stage ofdevelopment. We see this approach as an instance of biologically plausible datadriven inductive bias through pre-training. We built a computational model thatmimics this development mechanism by pre-training different artificialconvolutional neural networks with simulated retinal wave images. The resultingfeatures of this biologically plausible pre-training closely match the V1features of the primate visual system. We show that the performance gain bypre-training with retinal waves is similar to a state-of-the art pre-trainingpipeline. Our framework contains the retinal wave generator, as well as atraining strategy, which can be a first step in a curriculum learning basedtraining diet for various models of development. We release code, data andtrained networks to build the basis for future work on visual development andbased on a curriculum learning approach including prenatal development tosupport studies of innate vs. learned properties of the primate visual system.An additional benefit of our pre-trained networks for neuroscience or computervision applications is the absence of biases inherited from datasets likeImageNet.

ReWaRD: Retinal Waves for Pre-Training Artificial Neural Networks Mimicking Real Prenatal Development