Fig. 6: Results for empirical games of AlphaGo and MuJoCo soccer dataset. | Nature Communications

Fig. 6: Results for empirical games of AlphaGo and MuJoCo soccer dataset.

From: Navigating the landscape of multiplayer games

Fig. 6

Note that as these are empirical games, strategies here correspond to trained AI agents. In AlphaGo, the strong transitive relationship between agents is revealed via our analysis. In MuJoCo soccer, more complex relations between similarly-performing agents are revealed in the clusters produced.

Back to article page