Disentangling the Predictive Variance of Deep Ensembles through the Neural Tangent Kernel

Open in new window