|
pulsatrix 1.1.0
|
EvaluateReconstruction()'s result. More...
#include <featurizer_metrics.hpp>
Public Attributes | |
| int64_t | inputs = 0 |
| double | mse = 0 |
| Mean squared error over every element. | |
| double | explained_variance = 0 |
1 - Σ|x - x̂|² / Σ|x - mean(x)|²: the share of the inputs' variance, about their mean, that the reconstructions keep. | |
| double | cosine = 0 |
| The mean cosine similarity of each input and its reconstruction. | |
| double | norm_ratio = 0 |
The mean of |x̂| / |x|: below 1 means codes are shrunk. | |
| double | l0 = 0 |
| Active features per input: blocks, for a block-sparse featurizer. | |
| double | dead_fraction = 0 |
| The share of features that fired on none of these inputs. | |
| double | dense_fraction = 0 |
| The share that fired on more than the dense rate of them. | |
| std::vector< double > | firing_rates |
| Each feature's (or block's) share of inputs it fired on. | |
EvaluateReconstruction()'s result.
| double pulsatrix::ReconstructionMetrics::cosine = 0 |
The mean cosine similarity of each input and its reconstruction.
| double pulsatrix::ReconstructionMetrics::dead_fraction = 0 |
The share of features that fired on none of these inputs.
| double pulsatrix::ReconstructionMetrics::dense_fraction = 0 |
The share that fired on more than the dense rate of them.
| double pulsatrix::ReconstructionMetrics::explained_variance = 0 |
1 - Σ|x - x̂|² / Σ|x - mean(x)|²: the share of the inputs' variance, about their mean, that the reconstructions keep.
| std::vector<double> pulsatrix::ReconstructionMetrics::firing_rates |
Each feature's (or block's) share of inputs it fired on.
| int64_t pulsatrix::ReconstructionMetrics::inputs = 0 |
| double pulsatrix::ReconstructionMetrics::l0 = 0 |
Active features per input: blocks, for a block-sparse featurizer.
| double pulsatrix::ReconstructionMetrics::mse = 0 |
Mean squared error over every element.
| double pulsatrix::ReconstructionMetrics::norm_ratio = 0 |
The mean of |x̂| / |x|: below 1 means codes are shrunk.