Getting Started with Machine Learning: An Amateur’s Perspective
Machine Learning is a hot topic in technology today. Every company is moving towards adopting this technological revolution in one way or the other. There are a huge number of jobs being generated for people, with even a moderate machine learning expertise. All this seems very intriguing to a student, who is aiming to build a good technical career. But with only a few technological institutions including ML in their core curriculum, the student is left wondering “How do I begin Machine Learning? Should I even begin?”. With my own and several others’ experiences, this post aims to solve these and a lot many questions regarding an amateur approach to Machine Learning.
Firstly, for those who stumbled upon this post by mistake or with no idea about Machine Learning, let’s understand it through some simple examples you might encounter daily. The best example is that of the Google search. It’s so awesome when you write just a word, if not only some letters, and the suggested pop down menu has your search query right there! Siri, do this and that. Alexa, play some music. So convenient, right? And how YouTube forces you to continue streaming videos because they have those perfect recommendations every time! All this exists, thanks to Machine Learning. The simplest definition of ML is that it is the process of training a machine or a program, with the help of data, to learn a task, so that it can automatically do it.
That’s enough of the boring information about ML. Now let’s get to the point. To be clear, this post does not explain the basics of Machine Learning. This post will help you overcome your doubts about starting to work with Machine Learning. For an easier understanding, this post has been formulated into answers to the most commonly asked questions, some of which you would have definitely encountered. Here we go!
Is ML worth the try?
Is that even a doubt, given the load of attention and headlines it attracts every single day? As per Indeed, Artificial Intelligence Engineers were the most sought after jobs in 2018–19 (with a whopping average salary of $142,858.57 in 2019). Ranging from the simplest flow of traffic to robotics, ML has conquered a number of domains out there! And this is not all, Artificial Intelligence is projected to create 2.3 highly paid million jobs by this year, according to a report from Gartner. Statistics aside, being a millennial who wishes to bring big changes in the world out there, this is your go-to option today! And alright, everything aside, let’s talk about it in practical terms. Most students are unclear about what will be the main domain they would like to work in their technical jobs. And that’s a pretty common ordeal. But without exploring some domains, you can’t know. There is just no shortcut. So yes, try it because you don’t know your passion and forte yet.
Any pre-requisites?
Machine Learning has become a fancy domain because it is surrounded by its hype and it is now being sought after by many people as their career roles. But the good part is that you do not need to be very sound in computer science to start working with ML. All you need to know is the basic syntax of Python (which you can easily learn online in less than a week if you have learned C or some other language) and how to build some basic logic for programs. Yes take a sigh of relief, you DO NOT need to be a hardcore coder!
How do I actually start?
There are a lot of opinions on this question, and a lot of experiences as well. We filtered out the two most useful approaches to begin with Machine Learning.
- Start with an online course on ML (Or a book if you prefer that more)
A lot of people opt for online courses on websites like Coursera, Udemy, Edx, etc. All these websites contain a course similar to titles such as “Machine Learning for Beginners” and they are almost all good (Also get a course certificate too). These courses teach you everything needed to start working with machine learning. You will get to be familiar with a lot of new terminologies such as dataset, regression, etc., how the ML process actually works, how ML models are created and implemented, various algorithms used, etc. But an important thing that needs to be done along with these courses, is hands-on practice. Try creating small codes out of whatever you learn, throughout the course duration. After completion, take up a small project, implementing a ML model for some purpose, from everything you learned in the course. If you are not sure, a lot of project ideas are available on the Internet to choose from.
2. Start a project directly. Learn with the flow.
Many a times, the online courses might feel like containing extra unnecessary content, or even a waste of time. A bitter truth might be that you are just lazy to complete the course. Whatever the reason may be, you can still start with ML. Just take up a ML-based project and make Google your new best friend. Make sure you start with a common project which has various versions available on the Internet. But don’t just blindly copy them. Search the meanings of every word or function you do not understand. Compare and analyze the available similar projects for their different approaches. Try creating a project combining the other available projects. Everything may seem gibberish in the beginning, but you will start understanding it all soon. Keep doing this until you are confident enough to build your own ML project from scratch.
Incomprehensible codes! What do I do?
Starting with any approach, you will definitely be persuaded to look for code snippets on the Internet. And you will easily find a lot of them too. People generally advise not to copy codes from the Internet. But know that it’s perfectly fine and you are not on any wrong track. This is the learning process for ML. But the worst part is that when you lookup for the codes, they might seem so abstruse, that you might want to get away from ML. Do not lose hope here. Be patient and try to read through the lines of code, and understand them with the help of documentations or Google search. Try playing around with that snippet by tweaking values. Stopping is not the answer. Anything else might be!
Competitive Coding or Machine Learning?
While many people out there consider these closely related, there is a lot to differentiate between the two! Additionally, not only do these differences apply to the field of ML alone, but to a variety of other topics. Ethically speaking, the former seeks out to satisfy personal glory, while the latter is primarily focused on solving problem statements using related technologies. Undoubtedly, ML does not provide rewards right after you start with your work, it can rather be quite taxing for the user. In the end, companies look forward to people who prove to be the best fit for vacancies; and while it is good to have a good hand at both, the master of one trade is better than the jack of none. And if you have explored enough, and are very sure about the domain that you want to work in, proceed with it.
I started, finally! Now what?
Well, you've conquered your first step (Hurray!!); but as bad as it seems, it is just a baby step towards what you will be having to do next. Projects and papers are as much important as anything else, and expertise at this domain will help you solve real-world problems with the most optimal solutions. A very prominent advantage of ML stands on the prospect that it will help you to contribute to some research work as well and also publish some research papers to strengthen your profile (highly recommended for students wanting to pursue higher studies). Continue with your work, there are a countless number of organizations and universities looking for students who have approached and solved problems with such technologies.
Bringing it to a close.
Machine Learning has been one of the most disruptive technologies so far, and it promises to be one of the biggest game-changers out there. There a surplus number of technologies looking to make a difference, and Machine Learning promises to be one of the biggest prospects from the ever-expanding pool. Good luck with your Machine Learning journey!
