Getting Started with Machine Learning: A Beginner's Guide

The word machine learning might instill wonders in your mind like mine and for the past few months, I was trying to unravel the mystery behind the black curtains. So let's get started and I will be taking you down through my experiences after I finally decided to learn something new.

Machine learning is one of the top disruptive technology for the future and with this good read you will be taking baby steps into this entirely fascinating world.

As always we need to get the basics right first and then only we can build upon that. The fundamental breakdown of the different areas connected with ML(abbreviation for Machine Learning) is provided in the below ven diagram.

I hope it makes some sense to you, even if don't worry I didn't understand anything when I saw it for the first time. It shows how ML is related to the other big crowd pullers like AI(Artificial intelligence) and DL(Deep Learning). Here we can take DL as a subpart of ML and together they both make a part of the bigger AI. We also have many other areas like NLP, Image Recognition, and Data Science which are totally beyond the scope of this discussion.

In the present-day world, ML is of great importance and we make use of it to solve very complex real-life problems. With the help of enhancing computational power, this area grows exponentially. We can find these technologies rooted in almost all sectors and sections of human life ranging from basic smartphones to big spaceships and health, agriculture, military (and many more).

ML is different from normal programming where a computer generates the output for certain input values according to the program that we feed onto it. But in ML the input and output for a specific scenario are given to the machine and it finds out valuable insights from that information, using which it trains the model based on many of the learning algorithms to predict outputs for an entirely new set of input data.

Let's go get some ice cream🍦

There are many algorithms to talk about, which we will be covering in future blogs. For this post, we are sticking to pure basics. Let me give you a scenario, consider you and 2 of your friends are going to buy ice cream from a shop and the cost of ice cream for different numbers of people are listed in a chart as given below

Including you, it requires you to buy 3 ice creams and that value is missing on the chart, the human mind can easily calculate the rate for 3 ice creams and we get Rs.30, this is because of the cognitive and mathematical ability that our mind posses or gains over our life, but computers or machines in itself do not have this ability or intelligence. So we have to make use of various methods which involve step-by-step learning of the scenario with the help of algorithms. Generally, we call it the Learning algorithm which enables the machine to learn and help us in prediction or classification.

In the above case we used a price chart initially which contains all the data we need, it becomes our Data set(we have divisions within this data set, just keep in mind). The scenario explained here is a simple case of machine learning techniques called Supervised learning. It is a type of learning algorithm in which the initial learning datasets are provided by a human for the machine(program) to learn. Once the model learns via a process called training (repetitive learning processes carried over the datasets) helps in predicting a future value that the machine doesn't know. This is mainly by forming certain relationships or patterns in the given input data.

TLDR; To summarize, here we have seen a simple example(which you may find strange) of supervised learning. ML is not required for solving this problem but in real life complex problems replace it. There are numerous learning techniques/methods and algorithms which we may cover in the future. Hope to write more and happy learning!