The neuronal receptive field (RF) is an expression applied to the space in which the presence of a stimulus changes the response of the same neuron.
The responses of visual neurons, as well as the phenomena of visual perception in general, are highly nonlinear functions of visual input (in mathematics, nonlinear systems are phenomena whose behavior cannot be expressed as the sum of the behaviors of its descriptors).
In contrast, the vision models used in science are based on the notion of a linear receptive field; in artificial intelligence and machine learning, because artificial neural networks are based on classical models of vision, they also use linear receptive fields.
“Vision modeling based on a linear receptive field presents several inherent problems: it changes with each input, assumes a set of basic functions for the visual system, and contradicts recent studies of dendritic calculations.”
Marcelo Bertalmio, first author of the study, University of Pompeii Fabra – Barcelona
The study was recently published in the journal Nature, Scientific reports. The paper proposes modeling the receptive field in a nonlinear way, introducing the concept of a substantially nonlinear receptive field or INRF
The paper proposes modeling the receptive field in a nonlinear way, introducing a substantially nonlinear receptive field or INRF. A study by Marcelo Bertalmio, Alex Gomez-Villa, Adrian Martin, Javier Vazquez-Coral and David Kane, researchers at UPF’s Department of Information and Communication Technology, and Jesus Malo, a researcher at the University of Valencia.
INRF, in addition to being physiologically plausible and embodying the principle of efficient representation, has a key property of broad implications: for several phenomena of vision science where linear RF must vary with input to predict responses, while RF is linear varies for each stimulus, the INRF can remain constant under different stimuli.
Bertalmio adds: “We have also proven that artificial neural networks with INRF modules instead of linear filters have incredibly improved performance and better mimic the basic human perception.” This research highlights the intrinsic nonlinear nature of receptive fields in vision and suggests a paradigm shift in both the science of vision and artificial intelligence.
University of Pompeii Fabra – Barcelona