Make your mouse move like a human - modern AI directly in web browser
summary
A tiny web playground records mouse trajectories between two checkboxes and trains a model in-browser with TensorFlow.js to predict the next cursor point from current and goal coordinates. About ~200 samples produce a model that reproduces human traits: sparse early points and a concentration of points near target as the cursor nears B. Implementation uses a simple two-layer sigmoid network trained in-browser; results mimic direction and end-stage jitter but remain imperfect and require a larger model and more data.
AI-written summary · about
chapters (3)
related talks
29mAdvanced UX with tensorflow.js
44mArtificial Intelligence and Machine Learning in the browser
24mThe next level apps for Web developers Machine learning on browsers 🤯
29mSuperpowers for next gen web apps: Machine learning
41mAI in Web: TensorFlow js
15mMachine learning for next gen web apps with TensorFlow.js
15mThe future of machine learning using JavaScript
32mUsing Machine Learning Models in JavaScript Application