Machine Learning in the browser
summary
Builds a working client-side ML app using TensorFlow.js and WebGL to run neural nets in the browser, generating synthetic RGB → complementary RGB pairs (normalizing 0–255) as training data. Implements dense layers (e.g. 64 → 32 → 16) with biases, uses SGD and mean squared error, tunes learning rate, batch size and epochs, and shows the model’s predictions becoming visually indistinguishable from the true complementary colors. Live demo runs end-to-end and completes successfully.
AI-written summary · about
chapters (7)
- 0:00Start
- 5:06CLIENT-SIDE ML
- 11:59CHECKLIST
- 13:09NEURAL NETWORK
- 16:00WEIGHTS
- 16:21DENSE LAYER
- 23:40OPTIMIZER
related talks
36mMachine Learning in the Browser
35mMachine Learning in the Browser with deeplearnJS
24mThe Future of Machine Learning & Javascript
29mMachine Learning with Deep Neural Networks in the Browser
48mDeep Learning in the Browser
34mA Powerful Introduction to Machine Learning with Js
41mPractical Machine Learning for Front-End Developers
34mA Powerful Introduction to Machine Learning with Js