首页|Patent Issued for Distributed application development platform (USPTO 11900084)
Patent Issued for Distributed application development platform (USPTO 11900084)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
From Alexandria, Virginia, NewsRx journalists report that a patent by the inventors Antiga, Luca (New York, NY, US), Chaton, Thomas Henri Marceau (New York, NY, US), Falcon, Williams (New York, NY, US), Walchli, Adrian (New York, NY, US), filed on May 1, 2023, was published online on February 13, 2024. The patent's assignee for patent number 11900084 is Grid.ai Inc. (New York, New York, United States). News editors obtained the following quote from the background information supplied by the inventors: "Artificial intelligence and/or machine learning (AI/ML) development is generally inaccessible to software developers. This is because AI/ML development not only requires an understanding of machine learning (e.g., training data processing, model architectures, output generation, etc.), but it also requires an understanding of distributed computing (since AI/ML training and deployment requires coordination between different processes running on different machines), process orchestration, microservice stacks (since different steps of the AI/ML pipeline use different services, such as experiment trackers, feature stores, output visualization, etc.), microservice integration, and other tools and disciplines. For example, video analysis may require coordination between: YouTube™, running on a first machine, that processes a set of videos; Scale™, running on a second machine, that provides a UI for video annotation; PyTorch Lightning™, running on a cluster of machines, that run multiple experiments and/or train the model; and StreamLit™, running on a third machine, that proof-of-concepts the user interface (e.g., displays a labelled video). A user would need to understand how to use YouTube™, PyTorch Lightning™, Scale™, and StreamLit™, and the associated APIs, in order to even begin approaching this task. Even after they understand how to use each service, they would have to understand how to connect the service outputs together. Even after they understand that, the user would need to understand how to split the work across each machine, provision each machine, coordinate processes between the machines, and generally manage machine usage.
NETL
NSTL
