An Adaptive Locally Connected Neuron Model: Focusing Neuron

TL;DR

This paper introduces a novel adaptive neuron model that learns its receptive field, enabling focus on informative inputs and improving performance over traditional dense neurons in various data sets.

Contribution

The paper proposes a new neuron model with learnable receptive fields using backpropagation, enhancing focus and performance in neural networks.

Findings

Focusing neurons can move receptive fields towards informative inputs.

Focusing layers outperform dense layers on synthetic and image data.

Focusing networks maintain performance even with significant weight pruning.

Abstract

This paper presents a new artificial neuron model capable of learning its receptive field in the topological domain of inputs. The model provides adaptive and differentiable local connectivity (plasticity) applicable to any domain. It requires no other tool than the backpropagation algorithm to learn its parameters which control the receptive field locations and apertures. This research explores whether this ability makes the neuron focus on informative inputs and yields any advantage over fully connected neurons. The experiments include tests of focusing neuron networks of one or two hidden layers on synthetic and well-known image recognition data sets. The results demonstrated that the focusing neurons can move their receptive fields towards more informative inputs. In the simple two-hidden layer networks, the focusing layers outperformed the dense layers in the classification of the 2D spatial data sets. Moreover, the focusing networks performed better than the dense networks even when 70$\%$ of the weights were pruned. The tests on convolutional networks revealed that using focusing layers instead of dense layers for the classification of convolutional features may work better in some data sets.