- Home
- AI & Machine Learning
AI & Machine Learning
How machines learn from data and make decisions: regression, neural networks, CNNs, clustering, search and game playing — the building blocks behind modern AI.
Gradient Descent
The algorithm that trains almost every AI model — walking downhill on a loss surface. Roll a ball across real 3D surfaces and tune the learning rate.
Linear Regression
The "hello world" of machine learning — fit the best straight line through data. See every error as a 3D square that shrinks as the model learns.
Logistic Regression
Predict yes/no probabilities with an S-shaped curve. Watch a 3D probability sheet bend to separate two classes as it trains.
Support Vector Machine (SVM)
Of all the lines that separate two classes, pick the one with the widest street between them. A kernel lifts data into 3D when no straight line works.
Neural Network (Forward Pass)
See exactly how a neural network turns inputs into a prediction — neuron by neuron, weight by weight — in a 3D network you can rotate.
Convolutional Neural Network (CNN)
How computers see images. Watch a 3×3 filter slide over a picture, build a feature map, then ReLU, max-pooling and flattening — all in 3D.
Transformers & Self-Attention
How ChatGPT-style models connect words. Every word asks a question (query), every word offers an answer (key), and softmax decides who to listen to.
K-Nearest Neighbours (KNN)
Classify a new point by asking its k closest neighbours to vote. See the distances, the neighbourhood bubble and the vote in 3D.
Naive Bayes Classifier
Count how often each word appears in spam and in normal mail, then use Bayes’ theorem to score a new message. Simple, fast and surprisingly accurate.
Decision Tree
A flowchart of yes/no questions learned from data. Watch each split appear as a wall on the floor while the tree grows above it.
K-Means Clustering
Let the computer discover groups in data by itself. Watch centroids hunt for the centre of each cluster in a 3D feature space.
Principal Component Analysis (PCA)
Reduce the number of features while keeping most of the information. Watch a 3D cloud get centred, rotated onto its principal axes and flattened to 2D.
A* Search
The path-finding algorithm behind games and maps. See how f = g + h focuses the search on the goal — and compare it with Dijkstra and greedy search on a 3D maze.
Minimax & Alpha–Beta Pruning
How a computer plays games like tic-tac-toe and chess. Watch values flow up a 3D game tree — and see alpha-beta cut away branches that can't matter.
Q-Learning (Reinforcement Learning)
An agent learns which action is best in every situation from rewards alone. Q-values spread backwards from the goal until a good path appears.
Perceptron
The simplest neural network: a single neuron that learns a straight line to separate two classes by fixing its mistakes one at a time.
Genetic Algorithm
Evolve a solution like nature does: keep the fittest, mix their genes, add random mutations and repeat until the target appears.
About AI & Machine Learning
Machine learning is maths that moves. Roll a ball down a loss surface with gradient descent, fit a regression line, bend a sigmoid surface, follow numbers through a neural network and a convolution filter, and watch k-means, KNN, decision trees and PCA work on 3D data. Search and game-playing AI — A* pathfinding and minimax with alpha-beta pruning — are here too. A perceptron and a genetic algorithm show two very different ways a machine can learn.