Machine Learning — An Introduction

Updated: Oct 30, 2020

Gain a solid understanding of Machine Learning, its algorithms, and use cases

At present day, the emerging field of Artificial Intelligence has become the biggest hype for the current generation. AI is a vast ocean in computer science and not only deals with mere things in computer science, but it covers a whole bunch of stuff like Image Processing, NLP, Summarization, Computer Vision, etc. To hail the whole concept of Artificial Intelligence, there has to be strong equipment to strengthen the base ideology of Statistics and Probability for accurate decisions. And here’s where computer scientists coined the term ‘Machine Learning’.


Every understanding starts with a definition. ‘Machine Learning is a subfield of computer science that gives computers the ability to learn without being explicitly programmed’. This definition is brought by Arthur Samuel who first coined the term Machine Learning while working at IBM. It states that Machine Learning is a subfield of computer science that can be used to train or teach a computer to learn itself without being programmed in a precise manner.


The first and foremost important advantage of Machine Learning (ML) is that it doesn’t need to be programmed in a highly efficient manner and they can be taught like how humans teach a four-year-old child. This advantage was loved by computer scientists and data scientists as they were able to manage their work and time efficiently. Another important advantage of ML is that they can improve their skills over time by powerful algorithms.

Types of Machine Learning

Machine Learning is broadly classified into two types based on the learning process of the model which are Supervised learning and Unsupervised learning.

(i) Supervised Learning:

For an easy understanding of this concept, we have to look directly at the words that make it up. ‘Supervised’ means ‘to observe’. So, in this process, we are supervising the machine learning models to be able to produce highly accurate results or decisions. To supervise the model, they are trained or taught with labeled dataset attributes. A simple example of Supervised learning would be predicting whether a person has a benign or malignant type of tumor with a cancer dataset that has labeled attributes.

The two primarily used algorithms for supervised learning are Classification and Regression algorithms. The classification type of learning is the process of predicting discrete class labels or categories (Ex: Predicting the benign or malignant type of tumor). Whereas, Regression type of learning is the process of predicting continuous values or numerical values (Ex: Predicting CO2 Emissions of cars).

(ii) Unsupervised Learning:

Unsupervised learning means that the model works on its own to discover information that may not visible to human eyes. This process deals with unlabeled or unknown data. Unsupervised learning has more difficult algorithms when compared to supervised learning algorithms since we know little or no information about the dataset. The most common and widely used algorithm for unsupervised learning is the Clustering algorithm. The Clustering algorithm finds patterns and divides the data point into groups that are similar to one and another. The groups are classified based on Structure, Summarization, and Anomaly Detection. The best example of unsupervised learning would be segmenting a bank’s customers based on certain characteristics.

Supervised vs Unsupervised Learning

In a supervised learning process, the algorithm is trained on a labeled dataset and produces the desired result. In contrast, the unsupervised learning process algorithms are trained on unlabeled or unknown datasets through which the model itself extracts patterns and makes sense of the unknown data. Supervised learning is a simpler method of programming whereas, Unsupervised learning is computationally complex. Algorithms used widely for supervised learning are Regression and Classification while the algorithm used for unsupervised is Clustering. Finally, we can expect higher accuracy results in Supervised learning than the results produced by Unsupervised learning.

Real-Life Examples of ML

At present, Machine Learning is everywhere. There are many solutions provided by Machine Learning which people can observe in their day-to-day life.

(i) Virtual Personal Assistants: Assist in finding information, when asked over voice. The best examples would be Siri, Alexa, and Google Now. For answering the given question, these systems recall your past queries, look for information, and finally, refines the best result for you. Later, these responses are stored for future preferences.

(ii) Recommendation Systems: Have you ever wondered how Netflix or Amazon recommends TV Shows or products? These recommendations are made by using Machine Learning. Personalized recommendations are made based on customer’s behavior whether it can be adding products to your cart on Amazon or your past watch history of films on Netflix.

(iii) Online Fraud Detection: Companies like PayPal are using Machine Learning to protect their customers’ money transactions against cyberattacks. The company uses a set of tools that helps them to compare millions of transactions and distinguish between legitimate or illegitimate transactions taking place between the buyers and sellers.

(iv) Virus Detection: Many healthcare organizations are shifting towards machine learning for the prediction of viruses. These systems assist doctors to detect whether a patient is affected by a virus or not. This is done, by training the machine learning model with a large number of data on the features of the virus.

Apart from the above-mentioned examples, there are many more to mention like search results refining, customer segmentation and the list goes on and on.

Final Thoughts!

Hope you enjoyed this article on Machine Learning. Instead of jumping directly to the math and coding side of Machine Learning, it is always better to start with some basics and foundations of machine learning. If you want to hold a tight grasp on the field of Machine Learning, it is necessary to keep yourself updated. Importantly, hands-on learning or the practical phase of Machine Learning is very important to keep your knowledge up to the level. You can also find great resources on the net and cool online courses to elevate your knowledge. Machine Learning might become a little frustrating when it comes to the probability phase but, never stop learning and rage your enthusiasm to learn more and more!

Happy Machine Learning!