Notes on machine learning
Gaussian processes
Blog series exploring Gaussian processes. Starts by building intuition with from-scratch implementations in Python, then fits a practical model with JAX and Optax on a real-world dataset.
Regression quattro stagioni
Linear regression parameter estimation implemented in four different ways: simple linear regression, ordinary least squares (OLS), gradient descent, and markov chain monte carlo (MCMC)
How to implement a neural network
Learn the basics of neural networks and how to implement them from scratch in Python. The tutorial starts with explaining gradient descent on the most basic models and goes along to explain hidden layers with non-linearities, backpropagation, and momentum. The math is explained along the way together with Python code examples.
How to implement a RNN
Learn how to implement a recurrent neural network (RNN) in Python with the help of NumPy. The tutorial explains the basics of backpropagation-through-time and discusses some of the difficulties of training recurrent networks.
Multi-armed bandit implementation
Learn the basics of Bayesian multi-armed bandit models while implementing a simple example in Python.
Cross-entropy classification
Introduction to the logistic and softmax classification functions and how to use them with the cross-entropy loss function.
Multivariate normal distribution: a primer
Introduction to the multivariate normal distribution that will show you how to represent, visualize, sample, and compute conditionals from this distribution.
TensorFlow Estimator & Dataset APIs
How to use TensorFlow's Estimator and Dataset APIs to train models.