Machine Learning ELI5
Have you ever saw a baby since their childhood and observed how they start walking?
Before babies can walk, they need to develop strength and coordination to maintain an upright posture independently. They also need to be able to bear most of their weight — at least momentarily — on one foot. So babies move closer to independent walking when they display these motor skills: Pulling oneself up into a standing position (by gripping furniture or holding onto someone), Walking with support, standing alone (momentarily). Walking for the first time is one of the most exciting and memorable milestones in the child’s development. Babies have been preparing to walk from an early age, and now all the rolling, sitting up, bottom shuffling, crawling, furniture cruising. Standing culminates in the baby’s newest adventure — first steps.
Now consider a computer or a machine that doesn’t know how to do a particular task (I don’t want to disrespect the machines, but let’s say that at this time, the device is dumb). It can’t walk (Robots), differentiate between an apple and a banana on its own. It needs some instructions on what an apple or banana is or needs some training on walking.
So, the basic idea to make a machine intelligent is to classify between different fruits or animals or flowers, walk, and do more tasks like a human is known as Artificial Intelligence. Then, it provides the data to a machine to learn from its experiences, known as Machine Learning.
Machine learning is the science of getting computers to act without being explicitly programmed.
Types of Machine Learning
In this blog, I’ll focus on mainly two types of Machine Learning:
- Supervised Learning: Imagine you’re a newbie in a Science Lab, and you experiment. After experimenting, you have got a supervisor who can tell you if you’re correct or wrong. After the advice, you either test the process again or improve your result or don’t do it.
This is an example of Supervised Machine Learning where you’ve got a supervisor to tell you if your output is right or wrong. Supervised Learning is the machine learning task of learning a function that maps an input based on example input-output pairs (labelled data). In Supervised Learning, you work with labelled data. In this type of Learning, we mainly focus on two ideas: Regression (where we have to predict a numerical/continuous variable output) and Classification (where we have to indicate which class this object belongs to!).
2. Unsupervised Learning: Imagine four people out there (two male and two female). One male and one female love to sing, and one male and one female love to exercise.
We can club them into two types. One being grouping according to their gender and another being according to their passion/hobby. There’s no supervision required to execute this task, and we can discover this pattern independently.
Unsupervised Learning is a machine learning task, which uses machine learning algorithms to analyze and cluster unlabeled data. Its ability to find similarities and differences in information make it the ideal solution for exploratory data analysis, cross-selling strategies, customer segmentation, and image recognition.
This blog was all about to give you a gist of what Machine Learning is like eli5. I’ll try to provide more information on Supervised Learning, Unsupervised Learning, and more by making different blogs on them. I will also come up with the implementation part.
