[D] How to represent time series data for an RNN where the sequence is a topological sorting of a DAG? For a given node, the model should be able to reference the node's ancestors in the DAG.

I'm working on building a model that can predict performance of agents that continually play two-player zero-sum games against each other. For a given agent pair that plays a game, the model should predict the result based on each involved agent's history.

Here's an example of the type of time series data I'm dealing with:

1. A plays B, A wins
2. A plays C, C wins
3. B plays C, B wins
4. D plays E, E wins

Assuming …

dag data machinelearning node reference rnn series sorting time time series