An interactive game looks at two core Machine Learning ideas: Classification (supervised learning) and Clustering (unsupervised learning)
Playing with a keyboard or screen reader
Use Tab / Shift+Tab to move between controls.
Within "Choose a mode" or "How many clusters", use the Arrow keys to change your choice, or Enter/Space on an option to pick it.
Choose a mode
Supervised Learning
This is classification: study a few labeled training examples per class, then predict the class of new, unseen emoji and check your accuracy.
Unsupervised Learning
This is clustering: find the structure yourself, sort the characters into clusters, and explain what feature connects each one.
How many clusters? (k)
You'll get 40 unlabeled Chinese characters as a raw, unlabeled dataset. Choose k, the number of clusters, drag characters into each one, then explain the rule your clustering follows.
3clusters
4clusters
5clusters
Playing with a keyboard or screen reader
Tab / Shift+Tab — move between the pool, each cluster's drop zone, and its title/notes field.
Arrow keys — while an item is focused, jump to the next/previous item in the same zone (or Home/End for the first/last). While a pool or drop zone itself is focused, jump to the next/previous zone.
Enter or Space on an item — pick it up (or put it back down if it's already picked up).
Enter or Space on the pool or a drop zone — move the picked‑up item there.
Escape — cancel picking up an item.
Items can also be dragged normally with a mouse or touch.
Unsupervised Learning
Unlabeled Dataset(40data points left to cluster)
Starting over will clear your current work and reset the dataset. Are you sure?