Deep learning

An Introduction to Deep Learning

Deep learning is a branch of artificial intelligence working by neural networks concept. Deep learning model concept is inspired from our human brain and the neural network model is the replica of the brain.

In Deep Learning, ‘deep’ refers to the many number of layers through which the data is transformed. Neural networks can be used to perform many different kind of tasks, such as clustering, classification or regression. Deeplearning model is able to learn without human supervision, and data that is both unstructured and unlabeled can be used.


The following diagram represents the basic structure of neural networks


neural network


The neural network architecture consists of several layers. The layers in the neural network identify patterns and predicts output. Each layer of the neural network is build by its previous layer. The first layer is called the input layer and the last layer is the output layer. Whenever human receive new information, our brain tries to compare it with already known or past objects. The same concept is implemented in deep neural networks.

Deep learning is a subset of Machine Learning. The major advantage of deep learning over machine learning is there is no need of feature extraction in deep learning. The neural networks combines the process of  feature extraction and classification.

The different types of Neural Neural Networks
  • Artificial Neural Networks (ANN)
  • Convolution Neural Networks (CNN)
  • Recurrent Neural Networks (RNN)
Artificial Neural Networks (ANN)

An ANN is a collection of connected units called artificial neurons (similar to biological neurons ). Each connection between neurons can transmit a signal to another neuron. These components functioning similar to the human brains and can be trained with deep learning algorithms. The different forms of data used in RNN are Tabular data, Image data, Text data.

Recurrent Neural Network (RNN)

RNN has a recurrent connection on the hidden state. i.e., the neural network has connection to its present state itself in addition to the next states. The different forms of data used in RNN are Text data, Time Series data, Audio data.

Convolution Neural Network (CNN)

The building blocks of CNN is kernels. Kernels are used to extract the relevant features from the input using the convolution operation. CNN were used to solve problems related to image data.

Applications of Deep Learning
  • Self-driving cars is one of the major applications of neural networks.

  • Another most popular usage of deep learning is voice search & voice-activated intelligent assistants.

  • Other applications includes Image Recognition, Recommendation systems, speech recognition, etc.,


Leave a Comment