Deep Learning in the Browser
summary
Live demos show models running client-side: an explorable loan-default visualization that exposes thresholds and business impact, a real-time sentiment inference demo that flags toxic comments as typed, and a browser-based transfer-learning image prototyping flow that trains on user images. Implementation uses WebGL-accelerated JS stacks (tensorflow.js / deeplearn.js / ml5), data tooling (Arrow, Facets), and Observable notebooks; trade-offs and fixes discussed include model shipping for inference, quantization to shrink weights, privacy/IP risk, and limited on-device training capacity.
AI-written summary · about
chapters (17)
- 0:00Start
- 1:55Learning Paradigm
- 2:21Augmenting Intelligence
- 2:38Human in the Loop
- 4:44In the Browser, Really?
- 5:43Immediate Accees Reduce Friction Wider Audience
- 8:02Webgl accelerated Learning Frameworks
- 10:44Explorable Explanations
- 11:14Discovery by Active Learning
- 12:07Algorithms Data Data-Models
- 27:49Visualisation Reactivity Immediacy
- 28:36Multi-Disciplinary Skills Needed
- 29:41Model Inference
- 29:58Bring your Deep Learning Model to the Data.
- 36:12Data vs. Model privacy
- 37:19Easier build for Web & Mobile
- 37:37How do we Build?
related talks
32mDeep Learning in JS
36mMachine Learning in the Browser
35mMachine Learning in the Browser with deeplearnJS
34mThe Browser, The Machine Learning and The Future we might not have imagined just less than a year ago
35mThe Future of JavaScript & Machine Learning
34mA Powerful Introduction to Machine Learning with Js
39mRun Deep Learning models in the browser with JavaScript and ConvNetJS
54mDeep Learning in JS