liveBook · Manning

Readout function

As explained in Sutskever's List

A readout function combines node representations into a fixed-size representation or prediction for an entire graph.

A readout function pools the node representations produced by message passing into one vector, allowing a graph neural network to make a prediction about the whole graph. section 7.3.2

From atoms to a molecular prediction

In a molecular graph, the readout pools the atom states into a single vector used to predict a graph-level property such as energy, dipole, or gap. section 7.3.2

How Sutskever's List covers it

How deep it goes
Explains pooling node states into one vector but does not specify a pooling operation.
Examples
Uses atom representations pooled to predict molecular energy, dipole, or gap.
What you can do after
Understand how pooled node states support predictions about an entire graph.
How it connects
Connects message passing neural networks, graph neural networks, and molecular property prediction.
Cover of Sutskever's List
liveBookmanning.com
Read Sutskever's List on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Sutskever's List. The section links point to the parts of the book it draws on.