What is Deep Learning? | HackThatCORE

Image Source: Towards Data Science

In Recent years, Deep Learning has gain a lots of attentions lately in the field of Compute Science and absolutely for a good reason. It gave a powerful impact over the fields of study like Computer Vision and Natural Language Processing. But actually What is it and How does it works. Let's uncover these two questions...
Deep Learning is a Machine Learning Technique by which a machine can learn the various features and tasks by itself by analyzing and observing the data. This data can be in any form. May it be in the form of Images, Videos, Text or be in Sound. It can be used to train machine to perfrom certain tasks and increase its intelligence to think about the world like we, humans think and act.
Let's understand its working by taking a simple example in which we have to train a machine to tell the categories of the images of the transport systems. In this example, we have three images, one is of a Car, second one is of a Truck and other one is of a Boat. We have to train a machine through the technique of Deep Learning and its sub-technique known as Convolutional Neural Network (we will talk about CNN in the further posts) to categorize these images among the given categories.
Firstly Deep Learning algorithm tell the machine about the various characteristics of each type of objects in the images.

After analyzing all the images related to the objects, the machine can predict the category oof the respective image.
The second example of Deep Learning is the training of the robot. The Algorithms train the microcontrollers that how to move its hand after sensing the object position in order to pull it.
Not at all, Deep learning putted its base in the field of Pattern recognitions for a long time. Deep Learning has been used in recognizing the hand-written postal codes since the 1990s. But these algorithmic techniques has become more popular over the last five years. This is only due to the three factors:

• Deep Learning algorithms are now become more accurate than peoples in predicting the results after analyzing the pattern over Convolutional and Artificial Neural Network.
• GPUs enable us to find deeper networks and more accurate patterns in very less time.
• Large amounts of label data is now accessible using Deep Learning Techniques.

The most suited network in the Deep Learning Technique is CNN. CNN is actually more suitable for working with image Data. Actually in all these neural networks, the term deep is used to refer the number of hidden layers in the Neural Network. Traditional Neural Networks has probably 2 to 3 hidden layers but the more diverse network like CNN can contain hidden layers upto 150.
Here I have told you the basic introduction about Deep Learning, The upcoming post of my blog clears your all concepts about the Deep Learning, Machine Learning and Artificial Intelligence.