GT-CHES: Graph transformation for classification in human evolutionary systems

What is it about?

We assume there will be data shift in human environments because humans are generally engaged in mixed-motive games that require constant adaptation. We pre-train our data so that the labels in the past are embedded into a different space by time period.

Featured Image

Why is it important?

If you assume your data will stay constant, or may drift, you will overfit. This is even more the case for data involving human interactions.