Topological Uncertainty: Monitoring trained neural networks through persistence of activation graphs

Open in new window