Luke Harold Miles


Current Work

I'm getting my master's degree in computer science at the University of Kentucky in Lexington.

I'm spending fall 2018 working for Brent Harrison, researching using natural language processing of exemplar stories (stories showing good and bad behavior) for agent alignment. This builds off of Harrison's previous work with Quixote and others' work on Cheherazade.

I'm on the board for the Lexington Traditional Dance Association. I'm learning how to run the sound system so I can help out more. I think contra dance is a glimmer of community in a time of atomization.

My email address is
If you would be interested in working with me on software or research (or if you just wanna say hi) then I'd love to meet and get pizza. I am interested in artificial intelligence, bioinformatics, and computer security. I have a pgp key, if you use pgp encryption.

Old Projects

I spent fall 2014 and spring 2015 working for Albert Jon Meier at WKU, trying to learn about ecosystems by applying the tools of linear algebra to matrices of food webs.

I spent fall 2015 and spring 2016 studying graph theory with Dominic Lanphier at Western Kentucky University. We investigated a measure of the strength or connectedness of graphs or networks known as integrity. Dr Lanphier and others had already established some bounds on the value I(G), and I attempted to establish narrower bounds on other classes of graphs.

In fall 2016, I learned about voting methods and social choice functions from the Handbook of Computational Social Choice, then used that to make a python software package for determining election winners from ranked ballots (using e.g. plurality, Copeland, or Simpson). The package was later significiantly improved and expanded by Bernardo Trevizan.

I spent spring 2017 working with Judy Goldsmith and Cory Siler helping them analyze political campaign messages, using the tools of computational complexity. The paper is on arxiv. In October 2017, I presented it at the Algorithmic Decision Theory Conference.

I spent summer 2017 developing a hedonic game simulator. I wrote a summary and put it on arxiv and I also presented the software at the Algorithmic Decision Theory Conference.

In spring 2018, I made a program for Thomas R Zentall which pigeons get trained and tested with. We are seeking to determine whether pigeons are capable of transitive inference. Like if X﹥Q and F﹥X, then will pigeons know that F﹥Q?

I spent summer 2018 working with my friend Max Williams helping Dane Morgan, Raphael Finkel, Ryan Jacobs, Mati Turner, and others develop their materials science machine learning tool MAST-ML. We hope it will help materials scientists to predict properties (e.g. melting point, energy capacity) of materials without having to actually synthesize them.


A few of my favorite people who also have websites (not linked above): Web accounts: Favorites:


I think thirty day projects (as outlined by UKY alumn Matt Cuts) are an amazing technique for creating and destroying habits. Some things I've each tried doing every day for a month:

I was fortunate to attend the Gatton Academy, a public residential STEM school in Bowling Green KY, for my junior and senior years of high school. I think it is an amazing institution with good culture for starting a career in science. If you have a child living in Kentucky interested in science, then I highly recommend you consider having them apply to Gatton. I met my girlfriend Luci Keller there.