ewoksndreg.intensities.torch_metrics.nmi_loss#
- ewoksndreg.intensities.torch_metrics.nmi_loss(source, target, bandwidth=0.01, reduction='none')[source]#
Calculates the Normalized Mutual Information between batch source and batch target
param source: Batch of Tensors param target: Batch of Tensors with same size as source param bandwidth: The size of the kernel function, determines the sample rate of the histogramm(default: 0.01) param reduction: If None: returns Tensor of NMI for every element in batch. If Sum or Mean: Applies operation to the resulting tensor
- Parameters:
source (
Tensor)target (
Tensor)bandwidth (
int)