Alibaba has announced it will open-source X-Deep Learning (XDL), the algorithm framework behind its marketing technology and big data platform Alimama. The source code and support documents’ release is slated for December.
XDL is the industry’s first deep learning framework for super-large-scale high-dimensional sparse data scenarios such as advertising, recommendation and search. It was developed by Alibaba’s marketing technology and big data platform Alimama based on its advertising business, and has been deployed at scale in demanding production scenarios such as this year’s “Singles Day” (Nov 11, China’s annual online shopping extravaganza).
Alibaba’s first open deep learning framework can complement existing frameworks such as TensorFlow, PyTorch and MXNet. Unlike these and other deep learning open source frameworks which are designed for low-dimensional dense data such as images, video, voice, etc, XDL’s advantage is that it is also a complete solution for high-dimensional sparse data scenarios.
Developers can easily implement the most advanced open source DL algorithms based on TensorFlow or PyTorch on the XDL framework. In addition, enterprises or individual users who are already using other open source frameworks can easily expand on the basis of the original system and enjoy the ultimate distributed capability of the high-dimensional sparse data scenario brought by XDL.
XDL’s high reference value is expected to help researchers and practitioners at all levels in the fields of recommendation, search, advertising, etc.
Author: Jessie Geng | Editor: Michael Sarazen