Machine Learning
Did you know that machines can learn? They can achieve it with Artificial Neural Networks.
Sometimes, we need help from computers in order to perform tasks that are either too complex or take too long to calculate. For example, when we need to predict stock market prices or recognize posts that can go viral. In order to perform such tasks, we provide machines/computers with the ability to “think”, “learn”, and “adapt”.
Machine learning attempts to simulate evolution and human neural networks (including the brain and nerves). Evolution simulators are known as genetic algorithms, and artificial brains as Artificial Neural Networks (ANN for short).
The term ‘neural’ comes from the word ‘neuron’, or nerve cell, which is the main component of a nervous tissue in our brain. The human brain inspired ANN models.
There are many types of ANN. The simplest one is called a “feed forward ANN”. This neural network has many “layers”, including an input layer, an output layer, and any number of hidden layers, usually one or two. In each layer, there are a number of nodes, which act like axons. Each node has a connection to every node in the next layer, and each connection has a weight. This weight is how strong the connection is.
Each number then goes through an activation function, which limits the output (usually either a sigmoid to keep the numbers between 0 and 1, a hyperbolic tangent to keep it between -1 and 1, or a comparison to 0.5 to keep it as 0 or 1).
Artificial Neural Networks Example
For example, we want cars (represented by rectangles in the video below) to move forwards and stay within the track. The first cars will not know what their goal is. Some won’t move, some will move forwards, while others will move backwards.
It will take a while until they ‘learn’ that they have to move forwards and stay in the middle of the road to survive. After many attempts, some cars can finally complete their task!
Nat’s Robotics Academy is preparing an exciting workshop that will teach you the basics of Genetic Algorithms and Artificial Neural Networks, and how to build them.
Stay tuned for our upcoming workshop dates!
Have a look at our current workshops.