A Wandering Mathematician personal notes

The Miracle of Data-Driven Modelling

A typical problem in machine learning is to fit a collection of labelled data points $(\bold{x}^1, y^1), \ldots, (\bold{x}^p, y^p)$ using some model function \[y = f(\bold{x}).\] Each point $\bold{x} = (x_1, \ldots, x_N)$ is a vector in $\R^N$, where $N$ could easily be about millions. For example, a black and white picture with a regular came... Read more