A few years ago I was sneak-peeking into ML/AI domain out of curiosity. Now, I am professionally working in the same domain. The journey was not tough but weird. It was all madness of going through ML courses, Google/Youtube, slithering through bulky statistics and math books, biting hard nuts of differential calculus and logarithms, working through Python and code libraries,…uff, after all this, the question is will the adventure like this make someone an ML expert?
What I learned from my ML venture is that all the above stuff can equip you with weapons but will not teach you how to fire them. You may be a master at working with data and an expert at applying algorithms and build ML models.
But what separates a great ML guy and a good one is the ability to smell the intelligence hidden in data jungle and more importantly, how to apply that intelligence for creating value in real life.
Training the Mind!
Though sounds philosophical it is true. Your mindset, if not trained, can be a big blocker on your way to becoming an ML expert, for that matter in any discipline. Okay, then how to train the mind for ML?… is it to be creative, innovative, suspicious, lazy thinking??… well, they definitely help, but to have your mind to think like an ML expert is entirely a different ball game.
More or less, we are trapped in “If-then-else” world. All our thinking and creativity revolves around it.
Can you write a usable program that doesn’t use an if condition in any form? … difficult, and you will feel handicapped without it.
On the other hand, machine learning and hence, their predictions defy human logic. If you find logic in it you can write a program for it and no need to have a machine to learn.
How can we beat this beast?… it is challenging, not impossible though. Below are some of the tips that should help you as a developer or technology expert to rewire your mind to align with the thinking that is required for a machine learning expert.
Focus on Problem Framing
Problem framing is a crucial step in any machine learning project. A solidly framed problem will make your project jump over 50% on the progress bar. I have not come across a course, pertaining to problem framing, as convincing as Google crash course on machine learning. This course emphasizes on problem framing and they have presented it very comprehensively. Go through the problem framing section of this course. Try to make rounds with this section several times so that it will help you to push the problem framing mindset deeper.
Zero Code Principle
Whenever you start on a new technology project, try to figure out if you can achieve the project without writing a single code. Though, It may seem impossible or challenging in most cases, give a hard shot at it, you may end up with pleasant surprises. Scrutinize the requirement and get into the heart of it, try to understand the ultimate objective of the requirement. This may throw you some clues to develop multiple approaches to fulfill the requirement with no or minimal coding. The takeaway from this exercise is it will break the mindset of thinking with default patterns and temptations to write codes.
Don’t Be Old Wine in a New Bottle
If you are building a new application or solution, follow the thumb rule of “old wine in a new bottle”. What does it mean is, if your current solution can be developed by using 10 years old technology then your solution is as old as the technology itself, an outdated solution. Try to look for the latest technologies or frameworks which are focussed on providing the kind of solutions you have been working on. (Future will be like multiple AI components talking to each other to achieve a much bigger objective in a much better way. All you have to do is identify what are the AI components required and how to combine them to talk to each other) This exercise will help you to come out of the trap that you tend to search for solutions within the limitations of your knowledge and abilities. One’s knowledge and abilities are always limited.
If your current solution can be developed by using 10 years old technology then your solution is as old as the technology itself, an outdated solution.
Dump the Pride of the Past
Dump the burden of carrying expertise of sunset technologies. You may be a veteran in a decade old technologies and you carry a lot of pride in it. However, it will not help you much beyond being proud. I was working incessantly on oracle forms and reports a decade back. I was sort of expert at it. Since 10 years, I never got a chance to work on them again. Nonetheless, hanging on this knowledge was looming over my imagination and approach to new projects as my thinking would start from those outdated technologies capability. Be brave, try to dump off those items that you feel have already disappeared or going to disappear soon. This will help you to relieve the burden of excess and make you efficient at learning new stuff with less conflict.
Keeping the mindset in line with change makes the change more enjoyable.