WHAT IS MACHINE LEARNING
Machine learning is a subset of Artificial Intelligence. It focuses on the development and improvement of computer systems that can adapt and learn without being explicitly programmed or without explicit instructions on what to do.
Machine learning is all around us right in front of our noses sometimes and I mean that literally (the device you are reading this article from is a testimony). Most of us see the term as a complex, nerdy thing. The term is actually straight forward as you will see.
For example, in self-driving cars, machine learning algorithms uses the sensors and cameras from the car to work effectively during driving. From the first drive, the car is already collecting data that would be used to predict and make subsequent rides smoother. The self-driving car teaches itself how to respond to the movements of other vehicles around it. If you ply a certain route often, it learns how to avoid inhibitions to smooth movements like potholes.
This is just one of the many uses of machine learning. Some others are;
· Google Translate
· Image recognition
· Getting feedback from social media
Methods of Machine Learning
There are methods of machine learning. The first two are the most widely adopted.
· Supervised machine learning
As the name implies, this method requires supervision to work. The supervision comes in form of past events that would be used to predict future events. For example, in banks they can predict customers that would repay back if they collect a loan using the historical data from other customers that collected loans.
· Unsupervised machine learning
In this instance, there is no historical data of some sort to work with. So, it works with the data at hand at tries to find a pattern or structure to the data. For example, when YouTube recommends music to you. At first, it does not have your music history, so it works with the ones that are already on your phone or it uses your location, age and other factors to recommend.
The others are;
· Reinforcement learning
This is basically the route everyone took in life. We learnt many things by trial and error, like learning how to write, walk, talk. This is also applied in machine learning in areas like gaming and robotics. After a while, through the trial and error the system knows which way gives the best rewards.
· Semi supervised learning
This is used in applications of supervised learning but as in all things, the finances matter and supervised learning is expensive (the gathering of data, the effort required etc.). So semi supervised learning is like a middle ground between supervised and unsupervised machine learning.
Is Programming Knowledge a must?
I get this question a lot and yes, programming knowledge is a must. Programming knowledge is an important aspect of machine learning. Someone can know all the fundamentals like probability, statistics, calculus, algebra but without knowing how to code, deploying models or even making the models in the first place would be nigh on impossible.
As a beginner is it better to understand the fundamentals of machine learning first or learn the coding first. This is highly debatable, though I’d go for learning the coding first.
On a side note, people involved in machine learning are called human beings or machine learning engineers (if you want more specificity, lol).
