The Glacius Blog
Worked explanations of the mathematics behind machine learning, with diagrams and calculations you can follow.
Follow via RSSMachine learning
Follow one training step by hand to see how backpropagation calculates gradients, how gradient descent uses them, and where the learning rate enters the calculation.
Read articleMore to explore
4 articlesMathematical intuition
Compare two neighbors by hand, then normalize the vectors to see exactly when cosine similarity and Euclidean distance produce the same ranking.
Read articleMathematical intuition
Dot product vs cosine similarity, explained visually. See how vector length changes a similarity score, with a worked example and the normalization formula.
Read articleMachine learning
Why gradient descent oscillates: a visual, worked example showing convergence, overshooting, and divergence as the learning rate changes.
Read article