Machine Learning with JavaScript
summary
Shows how to run models entirely in the browser using winkNLP, Transformers.js, and TensorFlow.js with concrete demos: wink English light model (~1 MB) that returns tokenization, POS, NER and readability stats; Transformers.js image-to-text via Hugging Face Spaces (1,200+ converted models available) using ONNX under the hood; a TensorFlow.js bot detector trained from a Kaggle set (50,000 rows reduced to 61 for demo) that reached ~90–100% train accuracy but produced 2 correct and 2 incorrect predictions on four test tweets. Deployment via Vercel/Spaces and Observable notebooks.
AI-written summary · about
chapters (7)
- 0:00Introduction and background
- 5:29Wink NLP library overview
- 16:07Transformers JS library demo
- 25:40TensorFlow JS capabilities
- 30:32Danfo JS for data analysis
- 39:37Building a bot detector model
- 55:27Q&A and closing remarks
related talks
39mMachine Learning Magic for Your JavaScript Application
37mMachine Learning Principles, with JavaScript?
18mTransformers.js: State-of-the-art Machine Learning for the web
25mTransformers.js: Building Next-Generation WebAI Applications
37mMachine Learning Principles, with JavaScript?
32mUsing Machine Learning Models in JavaScript Application
24mThe Future of Machine Learning & Javascript
46mArtificial Intelligence with JavaScript