The Web Neural Network API (WebNN) brings accelerated machine learning capabilities directly to web applications. With WebNN, developers can harness the power of neural networks within the browser environment, enabling a wide range of AI-driven use cases without relying on external servers or plugins.

What is WebNN?

WebNN is a JavaScript API that provides a high-level interface for executing neural network inference tasks efficiently on various hardware accelerators, such as CPUs, GPUs, and dedicated AI chips (sometimes called NPUs or TPUs). By utilizing hardware acceleration, WebNN enables faster and more power-efficient execution of machine learning models, making it ideal for real-time applications and scenarios where latency is critical.

In the WebNN API, the Operand objects represent input, output, and constant multi-dimensional arrays known as tensors. The NeuralNetworkContext defines a set of operations that facilitate the construction and execution of this computational graph. Such operations may be accelerated with dedicated hardware such as the GPUs, CPUs with extensions for deep learning, or dedicated ML accelerators. These operations defined by the WebNN API are required by models that address key application use cases. Additionally, the WebNN API provides affordances to builder a computational graph, compile the graph, execute the graph, and integrate the graph with other Web APIs that provide input data to the graph e.g. media APIs for image or video frames and sensor APIs for sensory data.

This example builds, compiles, and executes a graph comprised of three ops, takes four inputs and returns one output.

A core abstraction behind popular neural networks is a computational graph, a directed graph with its nodes corresponding to operations (ops) and input variables. One node’s output value is the input to another node.

The WebNN API brings this abstraction to the web.

In the WebNN API, the Operand objects represent input, output, and constant multi-dimensional arrays known as tensors. The NeuralNetworkContext defines a set of operations that facilitate the construction and execution of this computational graph. Such operations may be accelerated with dedicated hardware such as the GPUs, CPUs with extensions for deep learning, or dedicated ML accelerators. These operations defined by the WebNN API are required by models that address key application use cases.