What are algorithms?

Technically an algorithm is a set of instructions designed to perform a specific task or, in other words, a list of rules to follow in order to solve a problema. And that’s why we get used to find algorithms in everywhere dealing with multiple challenges. But, when we talk about algorithms, we mainly refer to a specific type called Machine Learning Algorithms, which are ones of the most applied since the last two decades.

Whereas traditional algorithms are, according to a mathematical and computational approach, sequences of steps programmed to achieve a goal that if you want to improve news steps have to be implemented, machine learning algorithms improve their performance automatically. They update themselves without the human addition of new steps or rules, so the strength of machine learning algorithms compared to traditional algorithms is that they consist of a statistical model that finds out from available data the most significant information for the prediction and can enhance its accuracy automatically whenever new data goes into.From Google Image Recognition algorithm to Amazon recommender algorithm, nowadays there is a great deal of algorithmic systems.  

Machine Learning Algorithms

Basically, a Machine Learning Algorithm is a statistical model that predicts something and improves its accuracy of prediction or classification independently of human intervention by iteratively updating the model with new data.​

Machine Learning in practice

Imagine, or play the video below, a group of scientists that gather one day with the intention of developing an accurate image classificator to save cats’ and dogs’ pictures in separate folders. If they go for a traditional algorithm, they must invent an infinitesimal set of rules like this:

–> if the animal is bigger than “x” –> size is a dog

–> if it is smaller than that “x” –> size is a cat

–> and so with countless possible physical features…

Instead, if they opt for a machine learning algorithm, they can choose a statistical model that synthesize well enough the data and gives a good prediction. Then, having fitted the model – we may suppose that scientists have chosen multilinear discrimination algorithm- with previously classified pictures –the training data-, the prediction accuracy of the model can continuously improve if the scientist program an automation that whenever there is new data available, the model has to be fitted again based on the previous and new data. 

Interested in our work?

You can collaborate with the project by sharing with us algorithms that are being implemented around you or by using the information in this directory to foster changes in your community.