Test Post (Perceptron Algorithm)

04 February, 2024

Categories: machine learning, perceptron, algorithm

My problems with this algorithm began when people online were suggesting that a step (piecewise constant) function is differentiable (including at points of discontinuity). “Just throw the gradient¹ descent optimisation algorithm at it!” is not gonna fly with me. I went in search of a more rigorous treatment of this algorithm.

\begin{equation} \dfrac{\partial}{\partial \theta_{i}} f(x, \Theta) \tag{1} \label{eqone} \end{equation}

This is a reference to the tagged equation \eqref{eqone}.