Creating algorithms is challenging and time-consuming, and that has encouraged research efforts into automatic machine learning (AutoML) systems that can simplify and democratize AI. Existing AutoML tools however cannot be applied to graphs. To address that deficiency, researchers from Tsinghua University have developed an AutoML framework and toolkit specifically designed for graph datasets and tasks.
AutoGL (Auto Graph Learning) is able to automatically handle all stages of graph learning problems, including dataset download & management, data preprocessing and feature engineering, model selection and training, hyperparameter tuning and ensembling. The researchers say AutoGL will reduce human labour as well as the biases in machine learning graph tasks on a large scale, and can also serve as a platform for users who want to implement and test their own auto or graph learning methods.
Graphs are everywhere. In a general sense, graphs are a powerful tool for representing rich and complex data produced by a variety of artificial and natural processes. Graphs can represent for example the molecular structure of a compound, protein interaction networks, and biological and biochemical associations in chemistry and material sciences. They are also widely used to represent people’s relationships in the social sciences domain.
A graph can be considered as a structured datatype that has nodes (entities that hold information) and edges (connections between nodes that also hold information) and thus has both a compositional nature and a relational nature. AutoGL supports the fully automatic machine learning of graph data on two of the most common tasks in graph-based machine learning: node classification and graph classification.
AutoGL requires Python version 3.6.0 and PyTorch 1.5.1 or newer. Different graph learning tasks are solved by different
AutoGL solvers, which make use of
auto feature engineer,
hyperparameter optimization, and
auto ensemble modules to automatically solve given tasks.
The following algorithms are currently supported in AutoGL:
The AutoGL project page is on the Tsinghua University website, and the code is on GitHub.
Reporter: Yuan Yuan | Editor: Michael Sarazen
Synced Report | A Survey of China’s Artificial Intelligence Solutions in Response to the COVID-19 Pandemic — 87 Case Studies from 700+ AI Vendors
This report offers a look at how China has leveraged artificial intelligence technologies in the battle against COVID-19. It is also available on Amazon Kindle. Along with this report, we also introduced a database covering additional 1428 artificial intelligence solutions from 12 pandemic scenarios.
Click here to find more reports from us.
We know you don’t want to miss any news or research breakthroughs. Subscribe to our popular newsletter Synced Global AI Weekly to get weekly AI updates.
wow, what a prety toolkit