Hands On Machine Learning Paradigms

Artificial Intelligence (AI) has a tremendous attention in recent years. John McCarthy who coined this term in 1956, defines it as “ The science and engineering of making intelligent machines”. AI is the intelligence demonstrated by machines, by performing smart tasks that typically require human intelligence.

The main core of AI is Machine learning (ML), Arthur Samuel said that ML is the “Field of study that gives computers the ability to learn without being explicitly programmed”, witch means that ML is the field of learning from experience, without the need of so much programming.

In this article, we will talk about ML concept and applications, presenting the three main paradigms : Supervised learning, Unsupervised learning, and Reinforcement learning.

let’s start by giving some examples of machine learning applications from our daily life :

Spam Detection : the famous example of ML is the spam detection application, where Spam filters detect unsolicited, unwanted, and virus-infested email (called spam) and stop it from getting into email inboxes using machine learning. Gmail uses spam filters to make sure they aren’t distributing spam messages. So, how does machine learning work for spam detection ?

The first step is training model, where we feed to the model (data) examples of spam messages and normal ones, so the model will learn the features of each email categorie and be able to detect the categorie of a new input, as shown below.

House pricing prediction : is another example of ML applications, in witch the model receives features of all houses and their prices as input data to learn the relationship between them and be able to predict and estimate the price of a new house. This application will help people to know the price of the house that they want to buy.

Customer clustering : the third application is about segmenting custmers of an e-commerce website, in order to understand custmers behaviors and categories based on their demography and buying behavior. So here the model tries to divide customers into groups (clusters) on common features and characteristics. So that the website can market to each group effectively and appropriately.

So now, it’s time to know what are Machine learning techniques and algorithms. Actually, there is three main paradigms :

Supervises learning : uses labeled data and performs a mapping of input function to output function to solve classification and regression problems

• Spam detestion = classification
• House pricing = regression

Unsupervised learning : where we use unlabeled data sets to learn the funtions that can be used for describing hidden patterns and structures to do clustering and dimential reduction.

• Custmer segmentation = clustering
• Reduce the numer of features input = dimentional reduction

Reinforcement learning : witch focues on making decisions that are generated by mapping the situations to actions and evaluating witch action have to be considred for maximising a long term reward.

• It can be applied for automatic parking policies or autonomous driving cars.

That was an overview about machine learning concept and techniques, hope it was helpful for you. In next stories, we will make an in-depth for supervised learning algorithms.