In this current world where everything is fast-paced and competent, Machine learning is something that assists for everything to happen as it is. You can see the applications of Machine learning everywhere nowadays, something that recognizes your face in unlocking a phone to an estimate of how long you can take to read this article which is mentioned in the top. So here I am just going to try to explain machine learning in simple words to someone who is very new to this word.
One of the best ways to explain how a machine learning algorithm works is to draw an analogy between you preparing for some exam and how the algorithm works. For example lets assume that you are preparing for an exam and reading a set of questions and learning answers for them, similarly a model trains itself by reading through the data that’s it has been given for. As you read for more number of hours you will become familiar with you read, but you should also know when to stop or else you will just remember what you read and might fail to answer the questions asked in the exam even if there is a minimal difference between the questions that you prepared for and the questions that has been asked in the exam. Now come back to the machine learning model where we will assume it has been asked to train for 100 iterations. Since it has read the data for more than required duration it's going to fail when you give a data to predict if the data has a lot of variations from the data that it trained for. This is termed as over-fitting the model, a model should not be over-fit or under-fit, i.e. you should neither read for more hours the same thing nor read the same for a very less number of hours. So now you might have a little understanding to what machine learning is, it is nothing but a machine preparing for its exam. Wait don’t go, lets see brief information on something called Deep Learning and created a revolution in this field some years ago. But to explain more on Deep learning you need to have a high-level understanding on Neural networks. Put in simple words, neural networks are similar to neurons in our body which transmits the message from the brain to all your organs. To have a better understanding on them, you can check out the article below,
But let me give you my understanding of Deep learning, again go back to our analogy where you are preparing for your exam. But this time you didn’t stop on just reading, you are putting more and deep effort i.e. assume you are writing notes and learning it. Similarly, assume the model that we used is a neural network model, and it had only two or three layers i.e (shallow network). But now we have increased the number of layers more to make it as a deep network. Now lets see what the catch since you took notes and prepared you might have reduced the time to prepare for the exam and might do the exam with a better performance. Similarly, the deep learning model now will train for fewer iterations and performs better when compared to older shallow network which trained for more iterations and performed bad. So lets wrap it for now, I believe that this article might have helped you to get a glimmer of what machine learning is. There are more terms in machine learning like parameters, hyper-parameter tuning, feature engineering etc, but what I have tried here is to create an analogy for machine learning and us humans. Let me end this with a quote from one of the pioneer in machine learning:
A lot of the progress in machine learning — and this is an unpopular opinion in academia — is driven by an increase in both computing power and data. An analogy is building a space rocket: You need a huge rocket engine, and you need a lot of fuel.
- Andrew NG
Thanks for reading this article, this is my first try in blogging so apologies for the language if I could not make you understand. I am also new to machine learning like you, lets understand and make use of it in a better way.