Blog

This is a blog post detailing Approximate Stein Classes for Truncated Density Estimation, by myself and my supervisor, Song Liu, which recently got accepted into ICML 2023. Introduction Pretend, for a moment, that you are the kind of person who likes to see where animals live, and you go out for the day to find where all the animal habitats are.

For those of you familiar with either HBO’s Game of Thrones, or George RR Martin’s A Song of Ice and Fire (ASOIAF), you will probably be aware of the vast differences between each character and the depth of each storyline.

This research serves as a pre-cursor to the work that I am carrying out as part of my PhD, and involves unnormalised and truncated probability density estimation. This is a field which I find extremely interesting, and I believe that it has far reaching applications not just across data science and machine learning, but across many other disciplines as well.

The notion of true random has perplexed scientists and statisticians for decades, if not longer. But there’s one process that we still don’t know much about - the human brain.

“2020 I’mma run the whole election” may not sound like the words of a lyrical genius, but I ask you to “name one genius that ain’t crazy”. For the duration of your read of this post, take a minute to separate an artist from their art.

The arrow assigment operator <- is useless. Before I’m crucified by the R community, hear me out and read this post. Every time I read code written by an academic, lecturer or someone who uses R frequently, I come across the arrow symbol <- used for assignment of variables.

In February I participated in a COMPASS hackathon, where me and my fellow students fit statistical models to try to improve predictions in forecasting electricity demand based on weather related variables.