“machine learning”

Estimates carbon emissions from training a machine learning model based on GPU, hours and region.

Machine-learning color palette generator that previews palettes on brand, website and illustration mockups.

Scroll-driven visual essay on fitting a line, loss, and gradient-based training.

Train a tiny neural network on toy datasets in the browser and watch the decision boundary form.

Move a classification threshold and see the ROC curve and AUC respond.

Visual essay on how bagging many trees makes a random forest.

Interactive walk-through of neurons, layers and activation in a small network.

Step-by-step interactive explanation of how a decision tree picks its splits.

Animated explanation of k-fold cross-validation and why it is used.

Interactive illustration of precision, recall and how they trade off.

Interactive demo of how model complexity trades bias against variance.

Interactive article showing how t-SNE settings change the maps it produces on toy data.

Plots pass/fail students by study versus social media hours and compares linear and non-linear SVM boundaries.

Editable training set of spam and normal messages; type a new message and see how Naive Bayes classifies it.

Add labeled points on a study-habits chart, set k, and see how a new point is classified by its nearest neighbors.

Sample apps and an expression playground for the WebNN API, which runs neural networks on device hardware from the browser.

Animated k-means run on a random point set, with data reset and speed control until the clusters converge.

Visual, interactive tutorial and sandbox for learning git commands and branching.

Three-screen lesson: a puzzle, an explorer where you change the step size, and a naming screen that explains the update rule.

Demos of neural networks trained directly in JavaScript for classification, regression and more.

Interactive article on gradient descent momentum with sliders that change the dynamics live.

Enlarges low-resolution images 2x or 4x with a machine-learning upscaler.

Fits a line to housing data by square footage, bedrooms or age, shows slope, intercept and R-squared, and predicts a price.

Suggests clean clip-art style shapes to replace your rough doodles as you draw.