A Neural Spiking Approach Compared to Deep Feedforward Networks on Stepwise Pixel Erasement

TL;DR

This paper compares a biologically plausible two-layer spiking neural network to deep convolutional neural networks in classifying MNIST images with stepwise pixel removal, demonstrating comparable accuracy and robustness.

Contribution

It introduces a biologically plausible two-layer spiking neural network trained with a learning rule and compares its performance to deep CNNs on occlusion robustness.

Findings

Spiking network achieves good accuracy on occluded MNIST images.

Spiking network demonstrates robustness comparable to deep CNNs.

Biologically plausible learning rules can be effective for robust object recognition.

Abstract

In real world scenarios, objects are often partially occluded. This requires a robustness for object recognition against these perturbations. Convolutional networks have shown good performances in classification tasks. The learned convolutional filters seem similar to receptive fields of simple cells found in the primary visual cortex. Alternatively, spiking neural networks are more biological plausible. We developed a two layer spiking network, trained on natural scenes with a biologically plausible learning rule. It is compared to two deep convolutional neural networks using a classification task of stepwise pixel erasement on MNIST. In comparison to these networks the spiking approach achieves good accuracy and robustness.