Learning Each Function with Machine Learning

Supervised Machine Learning, Unsupervised Machine Learning & Semi-Supervised Machine Learning

Machine Learning is a subset of Artificial Intelligence (AI) that provides computers with the ability to learn without being explicitly programmed and to take intelligent decisions. It also enables machines to grow and improve with experiences. There are 3 types of learning that are associated with Machine Learning & these are: supervised, unsupervised and semi-supervised learning.

Supervised: It works with the labeled data and the algorithms in it learn to predict the output from the input data itself.

Unsupervised: It works with the unlabeled data and the algorithms learn to inherent structure from the input data.

Semi-supervised: It has a mixed case; some data are labeled but most of it is unlabeled and a mixture of supervised and unsupervised techniques can be used collectively.

Supervised Machine Learning

Supervised learning is used by majority of the Machine Learning Practical uses.

Supervised learning is used when we have input variables (X) and an output variable (Y), random error term (positive or negative) with mean zero (ϵ), then we use an algorithm to learn the mapping function (f) from the input to the output.

Y = f(X) + ϵ

The goal is to approximate the mapping function so well that it would work for every new input data (x) and can predict the output variables (Y) for the same data.

Here the algorithm is all about learning from the training dataset and it as a teacher supervising the learning process.

There are two types of Supervised Learning Problems; Classification & Regression:

Classification: A classification problem is when the output variable is a category, such as “White” or “Black” or “Happy” and “not Happy”.

Regression: A regression problem is when the output variable is a real value, such as “Pound” or “Height”.

Unsupervised Machine Learning

Unsupervised learning is used when we only have input data (X) and no corresponding output variables (Y).

The main purpose for unsupervised learning is to model the underlying structure or distribution in the data to learn more about the data.

Since unlike Supervised Learning, there is no correct answer and there is no teacher to supervise, that is why it is called Unsupervised Learning. Algorithms depend upon their own devises to find and present the interesting structure in the data.

Alike Supervised Learning, Unsupervised learning problems can also be further divided into clustering and association problems.

Clustering: A clustering problem is where we want to find the innate groupings in the data, such as grouping customers by ordering pattern.

Association:  An association problem is where we want to find rules/patterns that defines large portions of the data, such as people/group that order X also tends to order Y.

Semi-Supervised Machine Learning

We face problems where we have a large amount of input data (X) present and only some of the data is labeled (Y). That is known as Semi-Supervised Machine Learning Problem

These types of problems lie in between both supervised and unsupervised learning.

A good example is a menu at the restaurant where only some of the items are labeled, (e.g. Coffee, Sandwich, Burger) and most of the items are unlabeled.

Talking about unlabeled data, now a day it is very cheap & easy to collect and store. Therefore, most of the real-world Machine Learning Problems falls under this category. In the other hand, it can be expensive or time-consuming to label data as it may require access to domain experts.

We can use the supervised learning techniques to make best assumptions for predicting the unlabeled data, feed those data back into the supervised learning algorithm as training data and can use the model to make predictions on new unseen data.

Contact Person: Amelia Smith

Email Id: machinelearning@enggconferences.com; ameliasmithml@gmail.com