Machine Learning Fundamentals Explained with Simple Examples
Before we move forward to what is machine learning? Let us know why computers exist?
Machines are a lot better and faster than humans at several tasks with which they outperform humans. We can control these machines and get them to do jobs by giving them a set of instructions that we call programming. We offer machines instructions in terms of programs to do tasks for us. Computers came into this world because they allow humans to do certain tasks fast. They can perform certain tasks simultaneously, so computers are replacing humans in many fields and services in today’s world.
In earlier days, places where thousands of workers used to work, today they are replaced by machines. For example, I am new to this place, and I wanted to search for the nearest shopping mall to my house. So I can pull out a map and measure the distance of each mall from my home using a ruler, or I can use my computer and tell him to find the nearest shopping mall among these three for me, obviously through proper programming. Instead of wasting 10 minutes to find out the nearest shopping mall, I can use a computer to do this for me.
Now, these machines or computers are good at things which are easy for us to describe to them.
For example,
If the route of A < B,
If the way of A < C
If the route of A < D,
Then A is the nearest shopping mall among all.
But can the machine tell whether the person is sad or happy?
Let us consider an example of a company that wanted to pick up human emotions based on their reviews about a certain movie or a web series. Can a computer do that? Technically yes, but as the tasks get complex, so does the programming. Just by looking at a dog picture, a computer can’t tell you that it’s a dog; it has four legs, two ears, a furry body. Because it is difficult for computers to do tasks which are harder for us to describe. So we hire humans to do these things that are harder for us to explain to the machine. We like machines to take care of easier parts, something that we can describe, and things that are harder to tell. We let humans do it, like being an artist.
Machine Learning
Then comes machine learning, which has many applications in the modern world, such as self-driving cars, speech recognition, language processing, recommender system, and many more. Machine Learning provides the ability to the computers to perform actions like humans by providing them instructions.
For example, today, you know how to bake a tasty pancake because you are baking it for the last three years, and you see the recipe very well. So due to a lot of practice over time, you can make it taste great. With machine learning, we can give a machine the ability to bake a pancake as tastier as you do. It could be done by providing the data of your few years of baking process with a recipe and each quantity mentioned. From the data you had provided, the machine can find patterns and train itself to bake a pancake as well as you do.
Machine Learning can be further classified into several branches based on use cases. Those are as follows:
Supervised Machine Learning
In supervised machine learning, you are provided with labeled data with the target variable that you wanted to predict. For example, if you are told to develop a model to predict the house prices based on past data. In this case, you have given the target variable: the cost of the house and other independent variables such as location, type of material, age of home, and other past data.
Unsupervised Machine Learning
In unsupervised machine learning, the target variable is unknown, so you have to develop different solutions depending upon the given data. For example, you can build a recommendation engine based on the past genres of movies your subscribers watch before.
Reinforcement Learning
Reinforcement learning treats you as an agent who is at state zero. Let us understand it with an example, suppose you visit a website for the first time, and the website has Google Adsense, it will show you ads based on your experience. If you click on the ad, your state status changes from zero to one, and you will get a reward. It means Adsense will show you similar ads in the future. To summarize the whole thing, reinforcement learning learns from the environment by providing a reward or no reward.
This concludes our blog, and I hope you all enjoy reading it. Follow for more such content in the future.
