0 minutes read

Machine Learning: Everything you need to know in simple terms

Editorial Team
07/10/2019 4:52 PM

Is everyone around you talking about machine learning? They could be singing algorithms, techniques, models, metrics and you are wondering as to how all these terms fit into the picture. This is a good runway for you to have a perfect start…

A subject of broad and current interest

Machines have stood the test of time by keeping up with the ever-increasing demands of data processing. More and more researchers are constantly acknowledging the power and potential that machine learning has on a variety of fields. 

Machine learning is a relatively new subject, but not completely new for data scientists. If you want to confirm this, talk to statisticians, engineers, programmers, and mathematicians about this subject.  This is because the algorithms and other methodologies used in this machine learning today have been in use for years before even the term machine learning was established. Be rest assured that it’s not something that is still undergoing invention.

In this post, I will be sharing some interesting bits of this topic so that at the end of this post you are able to start making use of the vast techniques offered in the extremely interesting world of machine learning.

What is machine learning?

Think of how humans learn to speak. As they grow, they are exposed to examples of words and expressions that are already existing. It is from these expressions that an individual will continue to build to his or her vocabulary to an extent that he can figure out how to construct new more meaningful conversations. That is exactly how machine learning works.  In machine learning, we let a machine learn from examples for it to figure out how to respond to completely new input. Do you like fruit? Here is another example that can clear up any traces of confusion

First of all, we are trying to teach a machine to figure out what is the name of a fruit, given that we tell it a few features of a fruit. In this case let’s just work with a few features (color, shape). We can show the machine a fruit that is orange in color and circular in shape, then proceed to tell the machine that fruit with such properties is an orange. Let us continue to bombard our machine with more examples. I am going to tabulate the examples below. Bear with me on the names of colors. I have used such names simply for argument's sake.

  • Orange(round, light-orange)
  • Apple(round, green)
  • Banana(curved, yellow)
  • Orange(round, orange)
  • Banana(curved, light-green)
  • Orange(round, light-orange)
  • Banana(curved, dark-yellow)
  • Orange(round, light-orange)
  • Apple(round, brown-copper)
  • Orange(round, light-orange)
  • Apple(round, red)

What we have been doing is known as training our machine using examples. In a sense, our machine is learning from the data that we have supplied to it. Now it’s time to find out if our machine has learned. This time I will only feed features to the machine and expect it to tell the name of fruit. Let do it.

  1. ????? (Round, Orange). In this instance, the machine will tell me that that is orange. Well done to our machine.
  2. ???? (Curved, Light-yellow). The machine will respond saying it is a banana. Wow! Our machine is doing great

And that is it. The machine has learned to predict the name of the fruit using it's color and shape. We have carried out what is called machine learning. How does it feel so far? Easy? Fascinating? Oh wait, there is one more thing, do not confuse it with what you might have heard before.

This sounds like what I have heard before!

Wait! Take it easy. Machine learning may sound a lot like other sciences that you might have come across. You could have some knowledge about Artificial Intelligence. AI is way broader than ML. AI dwells on mimicking human behavior such as the ability to reason, discover meaning, generalize, or learn from past experience. ML focuses learning on just a part of AI that focuses on building algorithms that adapt behavior based on empirical data (learn from past experience)

Real world problems that you can solve using ML

Machine learning is everywhere, even in places where we cannot imagine. Its techniques are being applied in fraud detection, face recognition, job grading, essay grading, market segmentation, and many other areas.

Want to get started?

Check out this Google Machine learning crash course, a self-study guide for aspiring machine learning practitioners. It features a series of lessons with video lectures, real-world case studies, and hands-on practice exercises. Google also provides a machine learning glossary, which defines general machine learning terms as well as terms that are specific to machine learning library developed by Google called TensorFlow.

That’s it. Feel free to share with your friends your understanding of machine learning. 

Tapiwa Gomo is a Consultant at Industrial Psychology Consultants (Pvt) Ltd, a management and human resources consulting firm. Phone +263 (242) 481946-48/481950 or email: tapiwa@ipcconsultants.com  or visit our website at www.ipcconsultants.com

Editorial Team

This article was written by one of the consultants at IPC

Latest Posts

Lets Talk

Whether you're looking for more information or you're ready to start a project, We are ready to help


170 Arcturus Road, Greendale, Harare, Zimbabwe

Sign Up For Newsletter

Receive articles and jobs straight to your inbox