In this video, we are going to learn about deep learning and neural network. Before going to the artificial neural network and deep learning, let us see what a neural network is. Neural network is a network which is connecting our brain and all the parts of our body. For example, you have tea cup in front of you and we are able to smell it and we are able to see it and we are able to hear the sound of putting the cup on the table. All these things are input to our brain. See, the nose is an input device; Ear is an input device and the smell you are getting through your nose is an input to your brain. These things are sent to the brain through the neurons and these neurons are connected. How many neurons are there? We don’t know. Millions of neurons are there.
It is connected and finally we are making the decision that this is a tea and this is a tea with rose smell. Okay? So, this neural network, we wanted to replicate it to improve the result or improve the output in the machine learning process. So this neural network – we are trying to replicate and it is having an input; it may be having one input or five input or more than that or thousands of input at that time. This will lead to the next layer and next layer and finally it is making the decision that this is a tea cup and it is very hot and it will be having the smell of rose.
So this neural network is having an input part of it, and you having an output and in between, layers of neurons are there. In the artificial neural network, these neurons are called as nodes. So we are having a set of nodes which receive the input from outside and it is passing through the layers of nodes and finally it is getting a output. This has to be implemented in the computer so that - the brain – how it is working, this computer will also work in such a way to identify that this is a cup of tea with the smell of rose.
Now we are going to learn the concept of deep learning. Already we have seen neural network. Now when it has become complicated, when the numerous inputs are given, and we are going to have a network, we need deep learning. For example, a driverless car we are going to have and it is going to drive the car through a road. When it sees the signal post, and red signal is there, the car stops.
And when green signal comes, it moves. When a person is coming in front of the car, it stops. When another car is coming, it takes a direction in which there is space for it to move. This needs lots of input and you have to make decisions based on so many inputs instantly. See, you have a neural network for the input – what is the position of the car, what are the positions of the human being, cyclist and a motorist – everything it has to get us input and what is the space available it is having the input and it has to have layers of neural network. Then it is going to make a decision to move forward. This is very complicated and it is having multiple layers of neural networks. When the car is making the decision to move, such a lot of nodes it has to pass and it has to make a decision. That is called deep learning and deep learning is a subset of AI.
