Waking Up to the Potential of Machine Learning


…”BEEP! BEEP! BEEP!”  Your alarm goes off and you start your morning routine, somehow managing you to clock in on time without being yelled at by your boss. But, I bet that this didn’t happen to you on your second day of work (yes, I said second day, because everybody arrives at work at least one hour early on their first day). On your second day you thought you had all the time in the world to shower, eat, and whatever else we do in the morning to give our colleagues the impression that we are “professional.” However, despite your plan, that did not happen. You were probably at least 15 minutes late. And little did you know, that this experience would serve as the perfect analogy for your introduction to machine learning…

garfield alarm clock

Some days, even the ole’ Garfield clock can’t get you out of bed…

Days went by and suddenly you started clocking in on time, avoiding more of these embarrassing moments when the boss calls you out in front of everyone else in the office for being 2 minutes late. Now it’s a rainy Sunday and you lay on your bed satisfied with your routine while thinking to yourself ‘Gosh, I sure am glad I am not late to work anymore… I feel really bad for the new intern’.

Well, what if I asked you how you managed to adjust your morning routine to fit your time schedule so you wouldn’t be late anymore?

If your response is along the lines of ‘2+2=4’ then you are wrong, and I’ll tell you why. Despite what you may have thought, the most naive answer to this question would be a plan to “wake up at 7am, take 40 minutes to get ready and 15 minutes to get to work, always manage to clock in before 8am”.

So then, how did you manage to not be late to work when something unexpected happened and interrupted your morning ritual? For instance, let’s say that when brushing your teeth you realized you forgot to shave half of your face, or you had to take the trash out because you forgot to do so the night before. Something happened during your routine that affected your time schedule and would have eventually caused your allotted ’40 minutes to get ready’ to increase, making you be late for work. What do you do then? 

Alas, somehow you still were not late for work, and spared yourself another embarrassing moment. This was because you distributed your time to adjust for these unexpected events. Did you succeed on adjusting your time schedule when such events occurred on your first day? First week? Second week? No, but eventually you did. You learned to adapt and adjust to the demands of your situation over time.

bowl of lucky charms

Contains all of the major food groups: pink hearts, yellow moons, orange stars…

Well my friend, this is the process of learning. And this is what Machine Learning is all about.

At this point you might be wondering what Machine Learning could possibly have to do with running out of Lucky Charms, or seeing a big scary spider blocking the doorway. And the answer to that is… well, nothing .The answer is nothing. And now you think I’ve just wasted several minutes of your life reading this, and an hour or two of mine writing it.

But, wait! Don’t go yet! What I described to you is actually a scenario where you, as a person, are constrained by something; in this case, it’s time. Time gives rise to multiple problems, and you, by distributing your time, managed to solve them and accomplish your task of clocking in in time and avoiding the stigma of being the lazy one who always oversleeps.

Computers too, face constraints and problems. Machine Learning, an area that combines the statistical and the artificial intelligence worlds, represents the entire process of learning intuitively. It revolves around the idea of humanizing machines by helping them to learn from their experiences. And, just like for you, machines don’t always get it right on the first day.

For you not to be late to work anymore took time because your brain learned from the first days how long each task would take you, and you learned which of your morning tasks could be stretched out or reduced according to specific unexpected events that happened or could happen. Your brain learned to manipulate tasks, both expected and unexpected, with the goal of optimizing time. The philosophy behind Machine Learning is the same; computers, without our intervention, learn from different types of past and current data, in turn becoming smarter and able to perform each chosen task in the most efficient way. Just like us, the computers remember and analyze past experiences in order to optimize their performance for current or future events, and each task learned provides more performance benefits for the future.

 

rubber ducky

print(“Quack!”)

So the next time your computer doesn’t automatically know what you want to name your saved file, or your phone’s autocorrect decides that you want to tell your friend how “ducking” excited you are for your weekend trip, don’t get mad. Like your morning routine, it’s all part of the process of learning. And machine learning, like cognitive learning, is a process that gets better over time with each task completed and the appropriate feedback applied.

And that is how, after a series of disappointing mornings, you remembered to pickup that extra box of Lucky Charms last time you were at the store.

 

 

Contact Stefano

Stefano Evangelista
Stefano Evangelista
sevangelista@scmconnections.com

Stefano has been a Senior Supply Chain analyst with SCM Connections since 2017.Born and raised in Italy, Stef's favorite thing about living in the US is the abundance of 24-hour grocery stores.