Deep Learning Important Basics

Why Deep Learning:

Deep learning is one of the methods by which we can overcome the challenges of feature extraction. These models are capable to focus on the right features by themselves, which requires very little guidance from the programmer. These models also partially solve the dimensionality problem. Whenever there is high dimensionality data it has a lot of features and a lot of predictor variables, we use this concept. Deep learning extract features on its own and understand which features are important in predicting your output.

What is Deep Learning?

Deep learning is collection of statistical machine learning techniques used to learn feature hierarchies based on the concept of artificial neural networks. Artificial neural networks which are works exactly like how our brain works.

How does Deep Learning work?

People started coming up with deep learning, their main aim was to re-engineer the human brain. This is basically inspired from our brain structure. It studies the basic unit of a brain called the brain cell or a neuron. These neurons are replicated in DL as artificial neurons, which are also called perceptron’s.

A perceptron receives multiple inputs and apply various transformations and functions provide us an output. Here we are feeding input data to the artificial neuron or a perceptron by applying various functions it will give output. This is just like our brain consists of multiple connected neurons called neural networks, we also build a network of artificial neurons called artificial neural networks.

Understanding Deep Learning:

We understand deep learning by recognizing an image using deep networks:

  1. First we have to do is we are going to pass the high dimensional data to the input layer. To match the dimensional of the input data, the input layer will contain multiple sub layers of perceptron’s. So that it consumes the entire input.
  2. Output received from the input layer contains patterns and will only be able to identify the edges of the images, based on contrast levels.
  3. This output will be fed to hidden layer 1. Where it will be able to identify facial features.
  4. The output from hidden layer 1 fed to hidden layer 2. Here it will be able to form entire faces.
  5. Finally, the output layer performs classification, based on the result that you would get from your previous layers.

Applications of Deep Learning:

  • It is using in self-driving cars. In self-driving cars it will capture the images around it. It will process that huge amount of data and then it will decide what action should it take and that will reduce the amount of accidents.
  • It is also using in voice controlled assistance. In this we can tell whatever we want to do it will search for us and display for us.
  • In automatic image caption generation also we are using this concept. In this whatever image that we upload that react in such a way that will generate the caption accordingly.

It is also using in automatic machine translation. In this we can convert English language into Spanish and Spanish to Hindi. We can convert one language to another language by using deep learning concept.