Platform for AI (PAI) is an all-in-one platform that supports various machine learning algorithms, mainstream machine learning frameworks, and visualized modeling methods. This topic describes the features of PAI.
Rich machine learning algorithms
The algorithms of PAI are tested by business within Alibaba Group for many years. PAI supports basic algorithms such as clustering and regression, and complex algorithms such as text analysis and feature engineering.
Compatibility with Alibaba Cloud services
Models trained by PAI are stored in MaxCompute and can be used together with other Alibaba Cloud services.
All-in-one machine learning platform
PAI allows you to streamline workflows, including data uploading, data preprocessing, feature engineering, model training, and model evaluation and publishing.
Mainstream deep learning frameworks
PAI supports mainstream deep learning frameworks such as TensorFlow, Caffe, and MXNet.
Visualized modeling methods
Machine Learning Designer of PAI encapsulates common machine learning algorithm components and provides visualized modeling. You can drag and drop components to create machine learning experiments.
Quick model deployment
PAI allows you to deploy models trained by Machine Learning Designer and Data Science Workshop (DSW) as Restful APIs. This way, you can use models by calling API operations to meet your requirements.
Quality technical support
If you encounter any issues when you use PAI, contact your account manager.