Since at least April 2017 I have been studying artificial intelligence. So far this has been an idle line of research since I have not found any practical use of artificial intelligence software, but it remains a fascinating subject. I’ve read a few books on the dangers of artificial intelligence but I have also explored it on a less theoretical level. You can play around with neural networks because there is a wealth of open source software available and plenty of tutorials.
Artificial intelligence is incredibly difficult to learn. You will need to understand the advanced math of linear algebra and vector calculus. I don’t even know what those branches of mathematics entail. But my approach has been to learn what I can in the process. So far I have advanced my knowledge of Python, mathematics, statistics, computer vision, and Natural Language Processing. Currently I am reading Doing Math with Python: Use Programming to Explore Algebra, Statistics, Calculus, and More! by Amit Saha. This book is helping me to bridge the gap between programming and basic mathematics. But the book doesn’t really get into advanced math, so much of the material does not pertain to artificial intelligence. In fact, artificial intelligence is not mentioned anywhere in the book. Still, it can’t hurt for a computer programmer to improve his math skills and my math skills were extremely weak.
Messing around with artificial intelligence can really strain your computer resources. So far I have had to add another 4 GB of memory to my system to improve my processing times. I’m using a 32 bit system at work which often proves to be a limitation. You really need a 64 bit system to use some libraries. Fortunately new hardware is becoming available for artificial intelligence hobbyists and researchers. I bought a Movidius Neural Compute Stick which is a USB stick that contains a Vision Processing Unit (VPU). Unfortunately this requires you to be running Ubuntu 16.04 and you need an USB 3.0 port for the stick. I have recently bought a new laptop to meet these requirements. I have also ordered the AWS DeepLens from Amazon which is a camera designed for experimenting with deep learning frameworks. I also have my eye on a Parallella single-board computer but that is designed more for massively parallel computing and not artificial intelligence. I am reluctant to invest too much money in artificial intelligence without some clear upside to my professional development.