Perceptron Learning and Classification in a Modeled Cortical Pyramidal Cell.

Toviah Moldwin,Idan Segev

doi:10.3389/fncom.2020.00033

Toviah Moldwin, Idan Segev

Open Access

https://doi.org/10.3389/fncom.2020.00033

Copy DOI

Abstract

The perceptron learning algorithm and its multiple-layer extension, the backpropagation algorithm, are the foundations of the present-day machine learning revolution. However, these algorithms utilize a highly simplified mathematical abstraction of a neuron; it is not clear to what extent real biophysical neurons with morphologically-extended non-linear dendritic trees and conductance-based synapses can realize perceptron-like learning. Here we implemented the perceptron learning algorithm in a realistic biophysical model of a layer 5 cortical pyramidal cell with a full complement of non-linear dendritic channels. We tested this biophysical perceptron (BP) on a classification task, where it needed to correctly binarily classify 100, 1,000, or 2,000 patterns, and a generalization task, where it was required to discriminate between two “noisy” patterns. We show that the BP performs these tasks with an accuracy comparable to that of the original perceptron, though the classification capacity of the apical tuft is somewhat limited. We concluded that cortical pyramidal neurons can act as powerful classification devices.

Highlights

There has been a long-standing debate within the neuroscience community about the existence of “grandmother neurons”—individual cells that code for high-level concepts such as a person’s grandmother
To implement the perceptron learning algorithm in a modeled layer 5 thick tufted pyramidal cell (L5PC) we distributed excitatory conductance-based AMPA and NMDA synapses on the detailed model developed by Hay et al (2011)
In the simulations described above, we have demonstrated that the perceptron learning algorithm can be implemented in a detailed biophysical model of L5 pyramidal cell with conductance-based synapses and active dendrites

Summary

Introduction

There has been a long-standing debate within the neuroscience community about the existence of “grandmother neurons”—individual cells that code for high-level concepts such as a person’s grandmother. One remarkable aspect of this finding is that different images of the same celebrity would elicit a response in these neurons even if the subject of the image was facing a different direction, wearing different clothes, or under different lighting conditions. The specificity of these MTL cells is invariant to certain transformations of the sensory stimulus. Regardless of whether this finding is evidence for grandmother cells or merely for sparse coding (Quiroga et al, 2008), it is apparent that individual neurons can be highly selective for a particular pattern of sensory input and possess a certain level of generalization ability, or “tolerance,” to differences in the input that do not change the essence of the sensory scene

Methods

Results

Conclusion