Neural Networks - Introduction

Published Feb 3, 2019

Deep learning is a sub-field of Machine Learning. It is an artifical intelligence function that imitates the workings of the human brain. It helps in decision making by creating patterns that help to deal with complex input/output mappings.

Traditional ML methods work when the output is a simple function of the inputs.

Deep learning works best when it comes to complex relations in inputs such as

  • detecting object in image,
  • playing a video game,
  • predicting stock prices,
  • diagnosing illnesses,etc.

What makes the deep learning go?
Neural Networks.

Neural Networks

The way brain works is when it is hit by a stimulus, neurons are fired. Neurons process information and pass it to the next neuron which process it and passes on and on. The brain then takes a specific action according to the output. Neural networks mimics it.

Feel free to skip the next paragraph

Neural network is based on collection of connected nodes that loosely models the neurons of brain. Inputs are passed to nodes which processes it and signal additional artificial neurons connected to it.

This is how neural network looks like.

Neural Network

Wait a minute!

The image above was put with the intention of scaring you. Though the image above is accurate, it is not that scary or hard. The whole thing can be thought of as the following example.

Some data is given.

Data Points

What we have to do is to draw a line that seperates the blue and red dots.

Data Points Seperated

This is what neural networks help us in accomplishing.

That's it!

Go back