A now very popular framework for the development of sophisticated ML models. In 2017, Google also launched Kubeflow , an open source project that aims to enable distributed machine learning for Kubernetes . The previous flagship in the area of ML-as-a-Service was the Google Cloud ML Engine , which was integrated into the. Google Cloud AI Platform in 2019 . With the AI platform, Google is trying to bring all assets under one roof.
Which will cover the entire spectrum
of ML services, including data preparation DB to Data training, tuning and provision of models. AI Hub acts as a central hub for discovering, sharing and deploying ML models and contains a collection of models based on popular frameworks such as Tensorflow , PyTorch , Keras , XGBoost and Scikit-learn and running in Kubeflow , on virtual servers, Jupyter Notebooks or via Google’s AI APIs. As with Amazon SageMaker, the entire
ML workflow is heavily code-based
and runs via Python scripts, with the Aero Leads corresponding advantages in terms of flexibility and disadvantages in terms of usability. Amazon SageMaker has a somewhat broader range of integrable frameworks with MXNet , Chainer and SparkML , but the latest version of Tensorflow with all new features is always available on the Google AI Platform , while Amazon SageMaker is often a few weeks behind. ML platforms in comparison Amazon SageMaker
