The following order is made in accordance with the speech video and PPT.
The main share around the following five areas:
- PAI Product Overview
- Upload custom algorithms
- Plus the number of eco-smart market
- AutoML2.0 automatic parameter adjustment
- AutoLearning automatic learning
A, PAI Introduction and development functions
1. PAI-Studio
In the architecture of FIG PAI, the lowermost frame is calculated, and the PAI data resources. PAI supports MaxCompute, OSS, HDFS and NAS and other data sources. Frame data based on a variety of computing resources, born out of the earliest forms of PAI: PAI-Studio-- visual modeling experimental platform, two hundred Studio includes a plurality of algorithms, data preprocessing covering, characterized in engineering, model training, evaluation of experimental flow forecast full link operation. The user can drag and drop experiments constructed in PAI-Studio, without complicated operations. In addition, PAI built Kunpeng computing framework that can support features ten billion, ten billion VLSI training matrix samples. In the initial setup process, PAI-S