Neural Network Deceived by Visual Illusions Like Humans

Do neural networks dream visual illusions?

This is the question studied by researchers at the Department of Information and Communication Technologies, led by Marcelo Bertalmío together with Jesús Malo, a researcher at the University of Valencia.

In all three cases the Sagrada Familia has the same color, but is seen differently by the colors that surround it. This is a visual illusion.

Credit: UPF

Aconvolutional neural network is a type of artificial neural network in which the neurons are organized into receptive fields

in a very similar way to neurons in the visual cortex of a biological brain. Today, convolutional neural networks (CNNs) are found in a variety of autonomous systems (for example, face detection and recognition, autonomous vehicles, etc.). This type of network is highly effective in many artificial vision tasks, such as in image segmentation and classification, along with many other applications.

Convolution networks were born inspired by the behavior of the human visual system, in particular in its basic structure formed by the concatenation of modules composed of a linear operation followed by a nonlinear operation. A study published in the advanced online edition of the journal Vision Research studies the phenomenon of visual illusions in convolution networks compared to their effect on human vision. A work by Alexander Gómez Vila , Adrian Martín , Javier Vázquez-Corral , Marcelo Bertalmío , members of the Department of Information and Communication Technologies ( DTIC ) and carried out with the participation of the researcherJesús Malo from the University of Valencia..

“Because of this connection of CNNs with our visual system, in this paper we wanted to see if convolutional networks suffer from similar problems to our visual system. Hence, we focused on visual illusions. Visual illusions are images that our brain perceives differently from how they actually are”, explains Gómez Vila, first author of the study.

In their study, the authors trained CNNs for simple tasks also performed by human vision, such as denoising and deblurring. What they observed is that these CNNs trained under these experimental conditions are also “deceived” by brightness and colour visual illusions in the same way that visual illusions deceive humans.

Furthermore, as Gómez Villa explains, “for our work we also analyse when such illusions cause responses in the network that are not as physically expected, but neither do they match with human perception”, that is to say, cases in which CNNs obtain a different optical illusion than the illusion that humans would perceive.

The results of this study are consistent with the long-standing hypothesis that considers low-level visual illusions as a by-product of the optimization to natural environments (that a human sees in their everyday). Meanwhile, these results highlight the limitations and differences between the human visual system and CNNs artificial neural networks.

Contacts and sources: 
Núria Perez
Universitat Pompeu Fabra (UPF) – Barcelona

Publication: “Color illusions also deceive CNNs for low-level vision tasks: Analysis and Implications”,  Alexander Gómez Vila, Adrian Martín, Javier Vázquez-Corral, Marcelo Bertalmío, Jesús Malo (2020), September, Vision Research , advanced online edition. https://doi.org/10.1016/j.visres.2020.07.010