An intro to AI / ML and Deep Learning

You have sure surely heard of the terms Artificial Intelligence, Machine Learning and Deep Learning. But what is the differences between them terms? They all just refer to robots, right 🤖?

No.

Artificial Intelligence, Machine Learning and Deep Learning

Artificial Intelligence is a broad area of giving superpowers to machines. It refers to programs being able to perform tasks that require human intelligence i.e. reason, learn, adapt.

Machine Learning is within Artificial Intelligence, and it refers to the actual ability of machines to learn and improve themselves from data observations.

Deep Leaning is the heart of AI and it refers to the actual technique and process of learning. It is a programming paradigm that allows programs to learn from observational data. Learning can be supervised, semi-supervised or unsupervised.

The techniques above provide solutions to problems in image recognition, speech recognition, and natural language processing.

How does the training work in practice?

My favorite example is Cats and Dogs.

Similarly to the way our brain works, if we have numerous photos labeled either ‘car’ or ‘dog’, we can ‘teach’ a program to classify the type of animal it sees.

Teaching a model to identify cats and dogs source

Once we have labeled data, then the machine will try to identify patterns of pixels that will help it guess if an image depicts a cat or a dog. The best part is, that if the program incorrectly classifies an image as that of a ‘dog’ when it’s actually of a ‘cat’, it adapts and it improves its learning.

Souce

“In the end, the patterns form a machine-learned model, such as a deep neural network, that can correctly identify dogs and cats and fire fighters and many, many other things”.

In other words, we’re going from using outdated rules and if statements that would look into the speed of a woman in order to ‘guess’ the activity she performs, to..

.. identifying patterns from the actual data.