Machine Learning from Scratch: Intro
Hi, I hope you enjoy my first post on Medium. I believe artificial intelligence has the potential to be the most important technology in the world. Ever since learning about the field, I have been fascinated with AI and neural networks in particular. I have wanted to understand the latest developments in AI and even be able to replicate them myself, but something has frustrated me.
Get Perrin Alaine-Sedano’s stories in your inbox
Join Medium for free to get updates from this writer.
I’ve read and followed many tutorials about how to build machine learning systems. Almost all of them use libraries like TensorFlow, Torch, Keras, or scikit-learn. These tutorials are often excellent at explaining how to get results using these libraries, but I’ve always been left feeling that I don’t really understand what’s going on under the hood. Now I’ve decided to fix that.
I’m going to write a series of posts where I implement a wide variety of machine learning systems using just Python and Numpy, without any ML libraries. I’m planning to start small and eventually work my way up to the cutting-edge. For each post, I’ll also go through the math and concepts and then show how they can be translated into code.
Someone else out there has probably done this already. If so, I’m certainly not trying to take credit for the idea. My goal in building these models from scratch is (1) to learn, and (2) to educate others. All the code will be posted on my github account and will be free for anyone to use. I’ll be glad if this series is fun and informative.
What I cannot create, I do not understand. — Richard P. Feynman
