Deep Learning is a subfield of Machine Learning. It has wide scope and it has various application like automated driving, object detection, healthcare and many more.
For example, by using deep learning we can create a system which can detect multiple objects in a frame and also label them accordingly.
Deep learning is not so new, it is fairly old subfield of machine learning. It has achieved so much in past few years, like from Near-human-level image classification to Handwriting transcription and Speech Recognition to Digital Assistants(like Alexa) and many more. Isn’t it cool? 😉
So, what really is deep learning? And how ‘Deep’ it is?
It may be possible that whenever you hear about Deep learning, you could think that we should have deeper understanding of a problem for solving it but that’s not true.
The term Deep Learning was introduced by Rina Dechter in 1986. “Deep” in Deep learning refers to the different layers from which our data is passed and each layer gives some meaningful output data which can be fed to its successive layer as a input. There may be hundreds or thousands of layers in Deep Learning model and it is called as “Neural Network”.
As we know, Human brain consist a number of neurons connected to each other which creates a complex structure or a network of neurons. Just like that Deep learning model contains a number of layers and each layer consist a number of nodes(or neurons) which are connected to the neurons of its adjacent neurons. But that does not mean that deep learning model or neural network works like a human brain. It has its different working. You can see in following diagram how deep learning model looks like:
As you can see in above figure, there is 1 input layer, 1 output layer and between them there is another layer called Hidden layer. Layers present between input layer and output layer are called Hidden Layers. There may be number of hidden layers in neural network. Data is fed into input layer which transform data according to its weights and passes to next layer, same task is perform by its successive layer and output layer gives us final output.
There are different neural networks used in Deep Learning, we will discuss all of them in detail in upcoming articles of this series.
Some widely used neural networks are
- Convolutional Neural Network.
- Recurrent Neural Network.
- Long Short-Term Memory. and many more..
To help data scientists for making deep learning models easily and training them, many frameworks or library are introduced. Following are few frameworks or library for Deep Learning.
It is first article of Deep Learning series, In upcoming article we are going to learn so much about Deep Learning and also will do some hands-on. If you enjoyed this blog post, share it with a friend and stay tuned for next article.