For Hobbyists & Makers · k-Nearest Neighbors
k-Nearest Neighbors for a convolutional net
Built for hobbyists & makers exploring it for fun. Play with real physics and math, no license and no setup — just open and tinker. Simulate a convolutional net live below — adjust the inputs and watch it respond, right in your browser.
k-Nearest NeighborsLive
classify by your neighbors
Controls
Presets
k-NN classifies a point by majority vote of its k nearest labeled neighbors. Small k gives jagged, overfit boundaries that chase noise; large k smooths them out but can blur real structure. It is the simplest possible classifier — no training at all. Educational tool.
Data Inspector
Neighbors5
Boundarybalanced
Governing equation
Reading this result: k = 5 sits in the sweet spot: enough neighbors to shrug off noise, few enough to keep the true boundary shape.
Runs locally in your browser — free forever. Scale to the cloud when reality gets heavy.
More with k-Nearest Neighbors
k-Nearest Neighbors for a neural networkk-Nearest Neighbors for a sorting algorithmk-Nearest Neighbors for a search treek-Nearest Neighbors for a hash tablek-Nearest Neighbors for a recommendation enginek-Nearest Neighbors for an image classifierk-Nearest Neighbors for a chatbotk-Nearest Neighbors for a pathfinding AI
Frequently asked questions
- Is this good for hobbyists & makers?
- Yes — this version of "k-Nearest Neighbors for a convolutional net" is framed for hobbyists & makers exploring it for fun. Play with real physics and math, no license and no setup — just open and tinker.
- Do I need to install anything?
- No. It runs in any modern browser, free, with no account required.