简体中文
EnglishWhat Is Web Machine Learning (webmachinelearning.github.io)?
Updated
Web Machine Learning (webmachinelearning.github.io) is the public site of the W3C Web Machine Learning Working Group and its sister Community Group. Its purpose is to make machine learning "a first-class web citizen" by standardizing Web APIs that let browsers run machine learning inference directly on the user's device. The site is a coordination and entry point for that standards work — it describes the mission, the benefits of in-browser inference, and the specifications the group is developing, and it invites people to join and contribute.
What the group actually does
The Working Group standardizes Web APIs for in-device machine learning inference. Two things are worth separating here:
- Inference, not training. The scope is running already-trained models in the browser, not building or training them there.
- Standardization, not a product. The group produces specifications within the W3C ecosystem; it does not ship a library or a hosted service.
The site describes its method as building on Community Group incubations as seeds: new proposals are discussed and explored in the sister Web Machine Learning Community Group before they move toward formal standardization. In other words, the Community Group is the front door for new ideas, and the Working Group is where those ideas become specifications.
The specifications it points to
The site names WebNN and WebMCP as examples of the web specifications the group works on, and uses "and more" to indicate the list is not exhaustive. If you are looking for the authoritative, current list of specifications and their status, treat the site as the starting point and follow through to the linked specification documents rather than relying on the summary text.
Why in-browser inference: the four stated benefits
The site frames the value of its approach as "combining the best of both worlds" — performant numerical computation plus the reach of the web. It lists four concrete benefits:
| Benefit | What the site says it means |
|---|---|
| Low latency | In-browser inference enables new use cases with local media sources |
| Privacy preserving | User data stays on-device, preserving user privacy |
| High availability | No reliance on the network after initial asset caching, enabling offline use cases |
| Low cost | Computing on client devices means no server farms are needed |
These are the group's stated motivations for the standardization effort, not measured benchmarks. Read them as design goals that explain why the APIs are being standardized, and evaluate any specific implementation against your own latency, privacy, and offline requirements.
Who this site is for
- Web developers who want to understand where browser-based ML APIs are heading and what problems they are meant to solve.
- Specification contributors looking for the right venue — the Community Group for new proposals and discussion, the Working Group for formal standardization.
- Organizations deciding whether to track or participate in the standards process.
If you only need to run a model in a browser today, this site will not give you a tutorial or a ready-made library; it tells you what the standards landscape looks like and where the work happens.
How to use it
- Read the mission and benefits on the landing page to confirm the scope matches your interest (inference in the browser, on-device).
- Follow the links to WebNN, WebMCP, and other specifications to check current status and technical detail — the landing page is a summary, not the specification itself.
- Choose your venue: the Community Group for incubating and discussing new proposals, the Working Group for standardization work.
- Use the "Join us" invitation on the page to find the participation path if you want to contribute.
A practical note: because the site is a standards hub, its content changes as proposals mature. Check the linked specification pages for the latest state rather than treating any summary here or on the landing page as final.