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

Deep convolutional neural networks (CNNs) have structures that are looselyrelated to that of the primate visual cortex. Surprisingly, when these networksare trained for object classification, the activity of their early,intermediate, and later layers becomes closely related to activity patterns incorresponding parts of the primate ventral visual stream. The activitystatistics are far from identical, but perhaps remaining differences can beminimized in order to produce artificial networks with highly brain-likeactivity and performance, which would provide a rich source of insight intoprimate vision. One way to align CNN activity more closely with neural activityis to add cost functions that directly drive deep layers to approximate neuralrecordings. However, suitably large datasets are particularly difficult toobtain for deep structures, such as the primate middle temporal area (MT). Towork around this barrier, we have developed a rich empirical model of activityin MT. The model is pixel-computable, so it can provide an arbitrarily large(but approximate) set of labels to better guide learning in the correspondinglayers of deep networks. Our model approximates a number of MT phenomena moreclosely than previous models. Furthermore, our model approximates populationstatistics in detail through fourteen parameter distributions that we estimatedfrom the electrophysiology literature. In general, deep networks with internalrepresentations that closely approximate those of the brain may help to clarifythe mechanisms that produce these representations, and the roles of variousproperties of these representations in performance of vision tasks. Althoughour empirical model inevitably differs from real neural activity, it allowstuning properties to be modulated independently, which may allow very detailedexploration of the origins and functional roles of these properties.

A Rich Source of Labels for Deep Network Models of the Primate Dorsal Visual Stream