# Post-synaptic potential regularization has potential

## TL;DR

This paper introduces post-synaptic potential regularization (PSP), a simple and effective method that improves neural network generalization, especially in deep architectures, outperforming traditional regularization techniques like $$ regularization.

## Contribution

The paper presents PSP, a novel regularization method that enhances generalization in deep neural networks, demonstrating competitive and improved performance over existing techniques.

## Key findings

- PSP achieves comparable accuracy to advanced methods on MNIST.
- PSP outperforms $$ regularization in CIFAR-10 deep models.
- Empirical results validate PSP's effectiveness across different scenarios.

## Abstract

Improving generalization is one of the main challenges for training deep neural networks on classification tasks. In particular, a number of techniques have been proposed, aiming to boost the performance on unseen data: from standard data augmentation techniques to the $\ell_2$ regularization, dropout, batch normalization, entropy-driven SGD and many more.\\ In this work we propose an elegant, simple and principled approach: post-synaptic potential regularization (PSP). We tested this regularization on a number of different state-of-the-art scenarios. Empirical results show that PSP achieves a classification error comparable to more sophisticated learning strategies in the MNIST scenario, while improves the generalization compared to $\ell_2$ regularization in deep architectures trained on CIFAR-10.