What is Deep Learning and Artificial Neural Networks?
As I explained in the previous post Introduction to Machine Learning, Machine Learning is a subset of Artificial Intelligence, that gives the computer the ability to learn without the need for sophisticated program codes.
How do we know if an animal is an elephant or a giraffe? As humans, since the day we are born, we see an elephant and a giraffe in real life, tv or the internet. So as time pass by, we now can spot the difference between an elephant and a giraffe. Elephants have large floppy ears, long trunk or nose, big body whereas a giraffe has long necks and legs, and are the tallest animals in the worlds. But how can we teach computers to spot the difference between two images (giraffes and elephants)? It would be good to have a representation for the trunk of the elephant, the large floppy ears, but trunks and ears themselves can be hard to detect, due to perspective distortions, shadows, angles etc.! The solution is to learn the representations as well. This is one of the great benefits of deep learning algorithms.
“Deep learning (also known as deep structured learning or hierarchical learning) is part of a broader family of machine learning methods based on learning data representations, as opposed to task-specific algorithms. Learning can be supervised, semi-supervised or unsupervised.”wiki
This brings us to Artificial Neural Networks.
Artificial Neural Networks (ANN)
In deep learning, we are trying to mimic our brain, how our brain cells work? Neurons are the fundamental unit of the brain which receives sensory input from the outside world through sensory neurons. Communicate with each other in electrical signals and chemical forms and then send information to muscles through the Motor neurons.
Artificial Neural Networks (ANN) a system or computational model that works similarly like the structure and functionality of the neurons in the human nervous system. As our human neurons, ANN has 3 different layers.
- Input layer similar to the dendrites of the human neuron. All inputs are fed to the input layer and send to the hidden layer.
- The hidden layer is similar to the cell body of the human neuron. The inputs received from the input layer is processed and sent to the output layer.
- Output layer similar to the exon of the human neuron. The received processed data will be shown at the output layer.