Visualizing and Understanding Convolutional Networks

This repository contains the implementation of the paper "Visualizing and Understanding Convolutional Networks". The paper introduces various techniques for visualizing and interpreting the inner workings of convolutional neural networks (CNNs).

Table of Contents

• Introduction
• Installation

Introduction

In this project, we aim to replicate the experiments and techniques described in the paper "Visualizing and Understanding Convolutional Networks" by Zeiler and Fergus. The paper provides insights into the inner workings of CNNs and proposes visualization techniques to understand the learned features and activations.

Installation

To run the code in this repository, you need to have the following dependencies installed:

• Python 3.x
• matplotlib
• numpy
• pytorch
• torchvision

To Do

• Visualizing the activations of the convolutional layers
• Visualizing filters of the convolutional layers