CNN – Convolutional Neural Networks

Share on facebook
Share on google
Share on twitter
Share on linkedin

In this tutorial, I’ll discuss Convolutional Neural Networks or (CNN). How they’re different from regular neural networks and how they’re used. Deep learning neural networks have empowered modern computing tasks. Providing breakthroughs in task processing that couldn’t really be performed without deep learning. One such breakthrough has been convolutional neural networks – or CNN’s – a special type of neural network.

CNN - Convolutional Neural Networks 1

What’s in it for you?

  1. Introduction to the CNN.
  2. What is a Convolution neural network?
  3. How CNN recognizes images?
  4. Layers in convolution neural network.
  5. Applications of CNNs.

Introduction to Convolution Neural Networks (CNN’s).

Convolutional Neural Networks, or CNN’s, are neural networks specialised to work with visual data, i.e. images and videos. Convolutional neural networks (CNN’s) have been very popular for their implementation in image and audio processing.

They are very similar to the vanilla neural networks. Each neuron in one layer is connected to every neuron in the next layer, they follow the same general principles of feedforward and backpropagation. However, there are certain features of CNN’s that make them perform extremely well on image processing tasks. A big challenge with typical neural networks is that they consider an entire input all at once.

CNN - Convolutional Neural Networks 2
Yann Lecun pioneer of Convolution Neural Network (CNN’s).

Yann LeCun is a French computer scientist and a Director of Facebook’s AI Research Group. He built the first Convolution Neural Network (CNN) call LeNet in 1988. LeNet was used for character recognization task.

What is a Convolution neural network?

convolutional neural network (CNN) is a deep, feed-forward artificial neural network in which the neural network preserves the hierarchical structure by learning internal feature representations and generalizing the features in the common image problems like object recognition and other computer vision problems. It is not restricted to images; it also achieves state-of-the-art results in natural language processing problems and speech recognition. A convolution neural network is also known in as ConvNet.

CNN - Convolutional Neural Networks 3
High-level general CNN architecture

How CNN recognizes images?

A big challenge with typical neural networks is that they consider an entire input all at once. For applications like image processing, for example, each of these three representations of the letter A would be treated as different inputs. Even though they’re just the same character represented in different locations or angles.

The solution to this involves preprocessing, which is powerful but limited in its implementation. Or convolutional input, which grabs small portions of an image and looks at just those.

Here’s how the convolutional neural network processes an image. Convolution input refers to overlapping functions and input. Using a filter to obtain a portion of an image, feeding what’s in the filter to a convolutional neural network neuron. Then the filter is moved, usually by a pixel. And the process is repeated until the entire image has been processed.

CNN - Convolutional Neural Networks 4

When discussing convolutional neural networks, we speak of pooling layers, which are interwoven with convolutional layers. They’re processing layers used for downsampling. If you’ve worked with image editing programs like Photoshop, then you’ll be familiar with downsampling which is the process of taking an image file and shrinking its size.

While that sounds pretty straightforward, a lot goes on in the background to shrink an image and make it look like the original at the end because downsampling requires that information, pixels, be removed from an image and you can’t just randomly choose a pixel of a specific color and remove it. You have to analyze the area around that pixel and understand what the area is trying to represent. The pooling layer simplifies the input by reducing the number of parameters.

Layers in convolution neural network.

Leave a Reply

Close Menu