Leaf Me Alone: Pattern Recognition and Machine Learning

UBUNTU 17.10 IS COMING OUT SO SOON! Short post today but I will do a longer post next week focusing on the beautiful new operating system that is getting released.

I am really sorry about the posting dry-spell of late. I have been immensely busy with school work and training. You could go so far as to describe my life with the following pseudocode:

> location = &Alex;
> self_approval = 0;
>
> if (location == school) || (location == judo);
>     self_approval = 1; // hooray!
>
> else
>     self_approval = 0; // GET BACK TO WORK

I finally have a course at school where I get to formally learn about Artificial Intelligence. My university has graduate level courses that are about AI for electrical engineers but until this year had no undergraduate courses for ECE students. The course is looking at pattern recognition and data processing and so far it has been fantastic.

AI has been an interest of mine for a while but I always have found it hard to get into. No matter how many question I answered the question pool always became exponentially larger. I discovered that having access to tools that simplify things for new users is really important. It allowed me to start to answer my own question rather than spending an eternity reading forums.

My first forray AI was using Google TensorFlow which is a great tool but not great for learning how to use machine learning for the first time. The concepts aren't too difficult but drowning in both syntax and concepts was a little overwhelming.

Understanding classifiers and other AI tools can be pretty tricky. A tool that has been immensely helpful has been PRTools which you can get here. This tool allows for really easy visualization of classifiers and data sets.

The courses initial plunge ended up giving the students a large data set relating to breast cancer data with 30 features and roughly 550 patients. It was by far the largest data set I have ever had to usewith around 17000 data points. After many hours of work we got a classifier together with decent accuracy (96% or so?) for detecting cancer based on the information given to us.

Second lab for the course had us dealing with leaves. We were to train a classifier to be able to identify 4 different kinds of leaves. I guess this classifier is already better than me because I don't any of those types of leaves.





Popular posts from this blog

Development of a Sentient Power Grid?

Python: My Anaconda Don't Want None

How to Install Ubuntu and Other Linux Distros