# Posts

- Visualising Hamiltonian Monte Carlo I made a visualisation of Hamiltonian (hybrid) Monte Carlo sampling for a 1-d harmonic potential.
- Modelling Bitcoin / Ethereum prices using Metcalfe's / Zipf's law A look at how well Metcalfe's law and Zipf's law model the price of both Bitcoin and Ethereum.
- Unsupervised domain adaption by backpropagation: method discussion and implementation In this post I'll explain Ganin et. al.'s "Unsupervised domain adaption by backpropagation" method for training deep neural networks in the presence of domain (or covariate) shift.
- A fast comparison of maximum likelihood estimation vs. maximum a-posteriori MLE vs MAP, in a simple example.
- Predicting the price of Magic: The Gathering cards after they are reprinted Using a recurrent neural network to forecast the price-drop (or draw-down) of Magic The Gathering cards when they are reprinted.
- Optimization and sampling methods Part 2: The harmonic oscillator This is part 2 in a short series of posts about methods for finding local minina, and on using molecular dynamics or Monte Carlo methods to sample from functions. Here I introduce the harmonic oscillator, an important physical model, and relate it to what we did in part 1. I'll derive Euler's method and show a simple NVE simulation.
- Optimization and sampling methods Part 1: Introducing gradient descent This is part 1 in a short series of posts about methods for finding local minina, and on using molecular dynamics or Monte Carlo methods to sample from functions. Here I will introduce gradient descent, with a simple convex optimization example and then briefly discuss the non-convex case.

subscribe via RSS