pulsatrix 1.1.0
Loading...
Searching...
No Matches
pulsatrix::UncertaintyDecomposition Struct Reference

DecomposeUncertainty()'s result: one value per input, entropies in nats. More...

#include <uncertainty.hpp>

Collaboration diagram for pulsatrix::UncertaintyDecomposition:

Public Attributes

Tensor mean
 The mean of the samples' probabilities, (N, K): the prediction to use.
 
std::vector< int64_t > predicted
 The most probable class of the mean, the lower index on a tie.
 
std::vector< double > total
 H[E p], the entropy of the mean.
 
std::vector< double > aleatoric
 E H[p], the samples' mean entropy.
 
std::vector< double > epistemic
 total - aleatoric, the mutual information (never negative).
 
std::vector< double > agreement
 The fraction of samples whose most probable class is the mean's.
 
int64_t samples = 0
 

Detailed Description

DecomposeUncertainty()'s result: one value per input, entropies in nats.

Member Data Documentation

◆ agreement

std::vector<double> pulsatrix::UncertaintyDecomposition::agreement

The fraction of samples whose most probable class is the mean's.

◆ aleatoric

std::vector<double> pulsatrix::UncertaintyDecomposition::aleatoric

E H[p], the samples' mean entropy.

◆ epistemic

std::vector<double> pulsatrix::UncertaintyDecomposition::epistemic

total - aleatoric, the mutual information (never negative).

◆ mean

Tensor pulsatrix::UncertaintyDecomposition::mean

The mean of the samples' probabilities, (N, K): the prediction to use.

◆ predicted

std::vector<int64_t> pulsatrix::UncertaintyDecomposition::predicted

The most probable class of the mean, the lower index on a tie.

◆ samples

int64_t pulsatrix::UncertaintyDecomposition::samples = 0

◆ total

std::vector<double> pulsatrix::UncertaintyDecomposition::total

H[E p], the entropy of the mean.


The documentation for this struct was generated from the following file: