Keynote: Building Great Web Experiences with AMP and TensorFlow.js

on YouTube ↗20m

" summary

Overview of practical tools and measured outcomes: AMP runtime plus AMP cache and the AMP optimizer aim to keep Largest Contentful Paint under the 2.5s threshold (over 4s = poor) and reduce First Input Delay below 100ms and Cumulative Layout Shift under 0.1. TensorFlow.js runs in-browser or on Node, ships pretrained models (COCO-SSD 90 classes, MediaPipe face mesh 468 keypoints, MoveNet 17, BlazePose 33), powers real-time features like toxicity filters, webcam pose physiotherapy and a QA model that extracts answers from text, and enables export from AutoML or Teach able Machine for client-side ML.

AI-written summary · about

" chapters (7)

1. 0:00Start
2. 1:17UX Pillars
3. 4:59Thresholds
4. 9:24Machine Learning is transforming web experiences
5. 11:34Optimizing page loading
6. 12:22TensorFlow.js is a library for machine learning in JavaScript
7. 18:45Training a custom model using transfer learning

" related talks

• 28mMachine Learning on the web with TensorFlow.jsJason Mayes, Sandeep Gupta, Ping Yu · Google I/O 2021
• 15mMachine learning for next gen web apps with TensorFlow.jsJason Mayes · Google I/O 2021
• 20mTensorFlow.js Bringing Machine Learning to the Web and BeyondNick Kreeger, Nikhil Thorat · JSConf EU 2019
• 29mSuperpowers for next gen web apps: Machine learningJason Mayes · Chrome Dev Summit 2020
• 39mMachine Learning Magic for Your JavaScript ApplicationYannick Assogba, Sandeep Gupta · Google I/O 2019
• 41mTensorFlow.js 101: ML in the Browser and BeyondJason Mayes · ML Conf EU 2020
• 41mAI in Web: TensorFlow jsGant Laborde · Devnexus 2022
• 15mThe future of machine learning using JavaScriptDoni Rubiagatra · CityJS Singapore 2023