Amazon has announced the release of a new project Neo-AI, which seeks to optimize the performance of machine learning models for a wide range of platforms. This work is based on SageMaker Neo, a platform to optimize ML models for specific deployment targets.
The motivation for this is to provide ML capabilities to edge devices that are often constrained with little to no connectivity to the cloud. Neo-AI compiles trained models for target devices to make them more optimized. Amazon SageMaker Neo can compile ML models built using popular frameworks and tools such as Apache MXNet, TensorFlow, and PyTorch.