Eccentricity dependent deep neural networks: Modeling invariance in human vision

Humans can recognize objects in a way that is invariant to scale, translation, and clutter. We use invariance theory as a conceptual basis, to computationally model this phenomenon. This theory discusses the role of eccentricity in human visual processing, and is a generalization of feedforward conv...

Full description

Bibliographic Details
Main Authors: Chen, Francis X., Roig Noguera, Gemma, Isik, Leyla, Boix Bosch, Xavier, Poggio, Tomaso A
Other Authors: Center for Brains, Minds, and Machines
Format: Article
Published: Association for the Advancement of Artificial Intelligence 2017
Online Access:http://hdl.handle.net/1721.1/112279
https://orcid.org/0000-0002-1909-257X
https://orcid.org/0000-0002-7470-0179
https://orcid.org/0000-0002-9255-0151
https://orcid.org/0000-0002-3944-0455
Description
Summary:Humans can recognize objects in a way that is invariant to scale, translation, and clutter. We use invariance theory as a conceptual basis, to computationally model this phenomenon. This theory discusses the role of eccentricity in human visual processing, and is a generalization of feedforward convolutional neural networks (CNNs). Our model explains some key psychophysical observations relating to invariant perception, while maintaining important similarities with biological neural architectures. To our knowledge, this work is the first to unify explanations of all three types of invariance, all while leveraging the power and neurological grounding of CNNs.