M2M Day 369: I have good news and bad news
This post is part of Month to Master, a 12-month accelerated learning project. For October, my goal is to defeat world champion Magnus Carlsen at a game of chess.
I have good news and bad news…
The good news is that I’m much more proficient at building, training, and deploying machine learning models. Today, in only 30 minutes, I was able to construct a brand new, more sophisticated machine learning model (a convolutional neural network) based around my chess data, upload the dataset and model to a cloud computer (via Floyd), and begin training the model on machine learning-optimized GPUs.
The bad new is that it doesn’t seem like this model is going to perform any better than my previous one. In other words, as anticipated, the problem with my chess algorithm is a function of how I’ve constructed my dataset.
The new model is still training, so I could wake up to a surprise, but I’m not counting on it.
With that said, it is interesting how much more quickly I’m able to build and deploy new machine learning code. In fact, everything I’ve done for this month’s challenge I could now likely complete in one day.
That’s the amazing and crazy thing about learning… By fighting through the unknown and the occasional frustration, something that previously took me 30 days now will only take me a few hours.
Even if I don’t succeed at defeating Magnus at a game of chess in the next few weeks, I’ve definitely leveled up my data science skills in a big way, which will serve me very well in the future on a number of project.
Read the next post. Read the previous post.
