Machine Learning Mastery
There is something addictive about running models. I can’t quite put my finger on it.
Maybe it is the engineer in me who likes to test hypotheses and wants to see the result of it. Or maybe I just like seeing GPUs run at my command. I’m not sure. But I really enjoy it.
I am definitely not an expert at machine and deep learning — I’ve been doing research in it for a bit more than a year now. It’s been fun, demoralizing, humbling, and exciting.
You realize that after running tens to hundreds of models with slight and/or large changes between the last and next one, that you never really know what to expect.
There is a mystery that you want to uncover. From these experiments, you start to build an “intuition” about what will work. But the reality is that you will never know. By just always guessing that something will not work, you’ll most likely be right.
And just like anything, when you get used to the tips and tricks, the places where you can get tripped up, you start to maneuver with more grace than when you first started. It’s a good feeling.
But I still get very intimidated by the new networks, the tens of papers that come out every day, and the unlimited number of posts made by experts that are light years ahead of me.
But from what I can tell, there is a genuine push to make this over-arching technology more accessible and hopefully democratized for the greater good of humanity. Sure, loads of people are going to extract value from what seems like a deep learning and computational revolution.
That doesn’t mean that you can’t have your place — where you learn at your own pace and push your own boundaries of knowledge discovery.
So there goes my rant that hints of the glimpse into the beauty of running machine learning models.
Thanks for reading.
