Machine Learning - csen2025.org

DBSCAN and density clustering showing noise.

Clusters of Any Shape, Plus a Category for Noise

I remember sitting in a windowless lab during my PhD, staring at a visualization of a dataset that looked like a spilled bag of salt, feeling a profound sense of betrayal. Every textbook I had read promised that clustering would…

Diagram showing hierarchical clustering tree structure.

A Tree of Clusters You Cut Wherever You Like

I remember sitting in a windowless lab during my PhD, staring at a dendrogram that looked less like a meaningful data structure and more like a tangled mess of Christmas lights. I had been taught that hierarchical clustering was this…

Diagram explaining clustering with k means.

K Means Needs You to Know K Before You Start

I remember sitting in a windowless lab during my postdoc, staring at a convergence plot that refused to settle, feeling that specific, dull ache of realizing the textbook explanation was essentially a lie. We are often taught that clustering with…

k nearest neighbours prediction error visualization

No Training at All, and a Very Expensive Prediction

I spent most of my PhD watching brilliant researchers build incredibly complex, multi-layered neural networks to solve problems that could have been handled by a simple distance metric. There is this pervasive, almost academic vanity that suggests if an algorithm…

Visualizing kernel methods intuition in high-dimensional space.

Comparing Points in a Space You Never Actually Build

I spent three months in a PhD lab trying to wrap my head around the “mathematical elegance” of Hilbert spaces, only to realize that most textbooks treat kernel methods intuition like a sacred, untouchable ritual. They throw high-dimensional geometry at…

Maximizing margins using support vector machines.

Finding the Widest Possible Gap Between Two Classes

I spent three years in academia watching people treat support vector machines like some sort of mystical, impenetrable black box that required a divine revelation to implement. I’ve sat through seminars where presenters used dense, intimidating notation to mask the…

XGBoost and LightGBM compared implementation diagram.

Two Implementations of the Same Idea With Different Priorities

I spent three months in a graduate lab once, chasing a 0.2% accuracy bump by blindly swapping libraries, only to realize I hadn’t actually understood why the loss curves were behaving so erratically. It’s a frustrating cycle: you see a…

Gradient boosting basics: trees correcting errors.

Each Tree Fixes What the Last One Got Wrong

I spent three years in academia watching people treat ensemble methods like a black box, throwing hyperparameter configurations at a wall to see what sticks. It’s frustrating to see most tutorials present gradient boosting basics as a series of impenetrable…

Diagram showing random forests explained visually.

Many Weak Opinions Averaged Beat One Confident Model

I spent three years in academia watching brilliant students treat machine learning models like black boxes, praying to the gods of hyperparameter tuning when their results diverged. I remember sitting in a windowless lab, staring at a convergence plot that…

Diagram showing decision trees and splits.

Every Split Asks One Question and Commits to It

I spent three years in academia watching brilliant researchers treat decision trees and splits as if they were some mystical, impenetrable black box. I’ve sat through seminars where people used ten-dollar words to describe a process that is, at its…