Direct Preference Optimization is all the rage now in LLMs, and rightly so! The derivation is neat (and very familiar to those experienced with reinforcement learning) and allows direct, preference-based finetuning of regression-trained LLMs without having to learn a reward model.
[Read More]
Hi there!
I’m a PhD student in the Department of Computing at Imperial College London. My research interests range from machine learning to quantitative analysis and modelling.
If you are interesting in knowing more about me or my research, then you are definitely in the right place.
Some Interesting Offline RL Methods (Early 2024)
Intro
[Read More]
An Intro to Offline Reinforcement Learning
What is Reinforcement Learning?
[Read More]
A Summary of Action Recognition So Far
This will be a post that is updated as I read and learn more about this topic. It is intended as a space where I can write and summarise what I think is the key material I’ve come across. Action recognition is a huge field and this page alone is...
[Read More]
Mathematical Investing 1 - Meandering with Markowitz
Using Linear Algebra to choose the best investments
This post, and indeed the successive posts in this series are focused on portfolio optimisation. Harry Markowitz is credited with developing the field of Modern Portfolio Theory, and won the Nobel Prize in Economics for developing the Markowitz Model (Minimum-Variance).
[Read More]
Relative (Cross-Sectional) Momentum Trading Strategy
Don't Stop Me Now
Another day, another post. This time it’s all about Momentum. Momentum is, as defined in the Oxford English Dictionary:
[Read More]
Starting out with Gaussian Processes
A Gaussian Process Primer
After taking the course, Mathematics for Machine Learning (MML), at university and reading the excellent book of the same name, I was well and truly hooked on Gaussian Processes. The MML book introduces the idea of applying the full Bayesian treatment to Linear Regression, leading to the simple and beautiful...
[Read More]
A Simple Trading Strategy
Setting the scene for Machine Learning in Quantitative Finance
We have all heard of Warren Buffet. The Oracle of Omaha, as of 2019, has achieved an average annual return of 20.5%. This far beats the average inflation rate of 9.95% per year since 1965. Compound interest means that $1 invested in 1965 is now worth over $23,000!
[Read More]