AI Apps

A live demonstration of K-Nearest Neighbor classification using a pre-trained 5-cluster monster dataset. Walk through 7 steps: training data, test monster design, centering, PCA projection to 3D, neighbor search, K-NN vote (K=3,4,5), and model retraining.

A step-by-step K-Means walkthrough using the 20-monster training set. Choose your X and Y axes, place three starting centroids, and watch the algorithm iterate assignment and update steps until convergence at k=3 clusters.

Enter a monster's features and walk through the complete K-Nearest Neighbor analysis step by step — center the data, project to a PCA map, find neighbors, and take the vote.

A classroom experiment in computational humor — fill in the same story twice using two different AI word-selection approaches, then rate the results.

Upload a handwritten digit and watch a single-layer CNN (convolution + pooling + dense) classify it. Explore filters, feature maps, and the full forward pass step by step.

Upload a handwritten digit and watch a 784→60→10 neural network classify it step by step. See every pixel value, every weighted sum, and every neuron activation — the full computation, transparent and tangible.

Train a simple perceptron to recognise handwritten 3s and 8s. Upload your own images or use the built-in demo dataset. Visualise attention maps, weight grids, and see exactly how a basic neural network learns.