W3C Workshop on Web and Machine Learning

🌱 A virtual event with pre-recorded talks and interactive sessions


Over the months of August and September 2020, W3C conducted a virtual workshop to pave the way for enriching the Open Web Platform with better foundations for machine learning.

Outcomes & Proceedings

34 presentations were recorded and published.

These presentations triggered 25 online workstreams which were discussed in 4 minuted live sessions.

The Workshop Report summarizes the key findings from the workshop.

What was the purpose of this workshop?

The primary goal of the workshop was to bring together providers of machine learning toolkits and framework providers with Web platform practitioners to enrich the Open Web Platform with better foundations for machine learning.

The secondary goals of the workshop were as follows:

  • Understand how machine learning fits into the Web technology stack,
  • Understand how browser-based machine learning fits into the machine learning ecosystem,
  • Explore the impact of machine learning technologies on Web browsers and Web applications,
  • Evaluate the opportunities for standardization around machine learning APIs and formats.

🧐 What topics were in scope?

The following topics were identified as in scope

  • Opportunities and Challenges of Browser Based Machine Learning
    • Privacy-First approach to machine learning
    • Real-time in-browser Machine Learning
    • Performance, compatibility, JS environment gaps
    • Domain-specific compilers for Machine Learning
  • Web Platform Foundations for Machine Learning
    • Web Platform: a 30,000 foot view
    • Web Platform and JS environment constraints
    • Bringing Machine Learning to the JS ecosystem with Machine Learning libraries
    • Accelerated graphics and compute APIs for Machine Learning
    • Fast, portable code with WebAssembly / WASI-nn
    • Access purpose-built Machine Learning hardware with WebNN
  • Machine Learning Experiences on the Web: A Developer's Perspective
    • On-device training in browser
    • Datasets on the Web & Schema.org vocabularies
    • Interoperability of Machine Learning models for the Web
    • High-level load & run model vs low-level graph builder API
    • Integration of models and in-browser data sources sensors, AV
    • Considerations when deploying models to the web
    • TensorFlow.js
    • ONNX.js
    • Magenta.js
    • ML5.js
    • Paddle.js
    • Machine Learning in Web Architecture
  • Machine Learning Experiences on the Web: A User's Perspective
    • Teachable Machine & Project Euphonia
    • Visualization of deep networks, "human-interpretable neural nets"
    • Web a11y opportunity
    • Cross-industry case studies
    • Media technologies roadmap for the Web
    • Enhancing media experiences with Machine Learning
    • Making art with Machine Learning
    • Making music with Machine Learning
    • Teaching machines how people speak
  • plenary Machine Learning Consensus Landscape
    • Who is doing what: what's happening in standards, what's happening in related open source projects.

🌐 What is W3C?

W3C is a voluntary standards consortium that convenes companies and communities to help structure productive discussions around existing and emerging technologies, and offers a Royalty-Free patent framework for Web Recommendations. We focus primarily on client-side (browser) technologies, and also have a mature history of vocabulary (or “ontology”) development. W3C develops work based on the priorities of our members and our community.

👋 Program Committee


  • Kelly Davis (Mozilla)
  • Anssi Kostiainen (Intel)


  • Göran Eriksson (Ericsson)
  • Dominique Hazaël-Massieux (W3C)
  • Ningxin Hu (Intel)
  • Dean Jackson (Apple)
  • Sangwhan Moon
  • Roy Ran (W3C)
  • Georg Rehm (DFKI)
  • Amy Siu (Beuth University of Applied Sciences, Berlin)
  • Nikhil Thorat (Google)