Home / Blog / Introduction to Deep Neural Networks
deep neural networks

Introduction to Deep Neural Networks

Here we discuss the introduction of deep neural networks with real-time examples for a better understanding. 

Basically, deep learning is implemented with the help of deep neural networks and deep networks are multiple hidden networks.

Different Layers of Deep Neural Networks

The first layer is the input layer after that some process happens and it will go to the next node or you can say to the hidden layer node. So, every node is interconnected. We have one more hidden layer where some functions will happen.

After this hidden layer where some functions can happen all the nodes are interconnected to each other after this hidden layer, there is an output layer in this output layer. We are going to check output whether it is equal to the desired output or not. If it is not we are going to update the weights. So, this is what a deep network looks like now there can be multiple hidden layers there can be hundreds of hidden layers.

deep neural network

But in machine learning, we are not able to process multiple hidden layers. So, because of deep learning, we have multiple hidden layers at once now.

 Let’s understand This with an Example:

deep neural network

We will take an image that has dour pixels as you can see the top two pixels are bright that is black in color and the bottom pixels are white in color. In the next step, we will divide these pixels and we will send these pixels to each and every node. So, we need four nodes to send each pixel, and then we provide random waits the white lines show the positive waits and the black lines show the negative waits.

The high brightness with black dots will consider negative when we see the next output of the next hidden layer it will be provided with the input with this particular layer. So, this will provide input with positive weights to this particular node and the second input will come from another particular node.

Negative weights show that the value has been negative which is shown by black dots. Positive values are shown by using white weights with gray dots all nodes in hidden layers show the inverse of that particular image. In this structure, we are getting negative values with positive weights which are negative, and negative value with negative weights which is positive so in the next layer we are getting something that is positive.

In the third layer, I want to take the inverse of that particular image for this I actually provide the negative weights to do when I provided the positive weight so it will stay where ever it is.

After all the procedures it will detect the output as we can see in the above image is a horizontal image. After that, we are going to calculate the difference between the actual output on the desired output and we are going to update the weights accordingly.

 Let’s understand This with an Example:

deep neural network

We have images here we provide the raw data to the first input layer. These input layers determine the patterns of local contrast and orally fixate those patterns of local contrasts. This means that it will differentiate on the basis of colors in luminosity and all those things so we will differentiate those things. After that in the next layer, it will determine the facial features like eyes, nose, ears, etc. Then it will activate those correct features for the correct face. Then it will be sent to the output layer you can add more hidden layers to solve more complex problems.

For example, if we find a particular face that has large eyes or which has light completion so we can do that by adding more hidden layers. And then we accumulate these features and find the output and determine the image. So in this way, a deep neural network looks like this.

Batch, Mini Batch & Stochastic Gradient Descent | What is Bias?

Simple Neural Network with a Single Neuron

Weights in Neural Network | Binary Classification