AI Model Local Deployment

Run models right in your browser with AI Squared!

To save cost and complexity, and frequently to comport with security requirements, the best way to make models available to end users in your organization would be to have those users run the model locally, right in the browser. In this set of examples, we show you how to use AI Squared to deploy neural networks in the browser on client devices with AI Squared.

Natural Language Processing:

Sentiment Analysis

Computer Vision

Image Classification

AI Squared supports neural network machine learning models built with TensorFlow because it automatically converts models from TensorFlow to TensorFlow.js to allow them to run in the browser. There are approaches to run e.g. scikit-learn models in the browser, for example with ONNX, and this is an active area of research and development. In the meantime, the aisquared python API contains a utility function (mimic_model) which will train a sparse neural network to mimic a scikit-learn model.