Fréchet Inception Distance
E290874
Fréchet Inception Distance is a widely used quantitative metric that measures the similarity between real and generated images by comparing their feature distributions extracted from a pretrained Inception network.
All labels observed (2)
| Label | Occurrences |
|---|---|
| Fréchet Inception Distance canonical | 4 |
| Fréchet Inception Score | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T2703887 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Fréchet Inception Distance Context triple: [Generative Adversarial Networks, evaluationMetric, Fréchet Inception Distance]
-
A.
Generative Adversarial Networks
Generative Adversarial Networks are a class of machine learning models in which two neural networks compete to generate highly realistic synthetic data, such as images, audio, or text.
-
B.
PixelRNN
PixelRNN is a deep generative model that uses recurrent neural networks to sequentially model and generate images pixel by pixel.
-
C.
PixelCNN
PixelCNN is a deep generative model that uses convolutional neural networks with autoregressive masking to model and generate images pixel by pixel.
-
D.
Automatic Adam
Automatic Adam is the nickname of Adam Vinatieri, a legendary NFL placekicker renowned for his clutch, game-winning field goals in high-pressure situations.
-
E.
Adam: A Method for Stochastic Optimization
"Adam: A Method for Stochastic Optimization" is a highly influential machine learning paper that introduces the Adam optimizer, a widely used adaptive gradient-based optimization algorithm for training deep neural networks.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: Fréchet Inception Distance Target entity description: Fréchet Inception Distance is a widely used quantitative metric that measures the similarity between real and generated images by comparing their feature distributions extracted from a pretrained Inception network.
-
A.
Generative Adversarial Networks
Generative Adversarial Networks are a class of machine learning models in which two neural networks compete to generate highly realistic synthetic data, such as images, audio, or text.
-
B.
PixelRNN
PixelRNN is a deep generative model that uses recurrent neural networks to sequentially model and generate images pixel by pixel.
-
C.
PixelCNN
PixelCNN is a deep generative model that uses convolutional neural networks with autoregressive masking to model and generate images pixel by pixel.
-
D.
Automatic Adam
Automatic Adam is the nickname of Adam Vinatieri, a legendary NFL placekicker renowned for his clutch, game-winning field goals in high-pressure situations.
-
E.
Adam: A Method for Stochastic Optimization
"Adam: A Method for Stochastic Optimization" is a highly influential machine learning paper that introduces the Adam optimizer, a widely used adaptive gradient-based optimization algorithm for training deep neural networks.
- F. None of above. chosen
Statements (49)
| Predicate | Object |
|---|---|
| instanceOf |
generative model evaluation metric
ⓘ
image quality metric ⓘ statistical distance ⓘ |
| alsoKnownAs |
FID
ⓘ
Fréchet Inception Distance ⓘ
surface form:
Fréchet Inception Score
|
| appliedTo |
GAN evaluation
ⓘ
generative adversarial networks ⓘ image generation models ⓘ |
| assumes | Gaussian distribution of deep features ⓘ |
| basedOn | Fréchet distance between multivariate Gaussians ⓘ |
| captures | both mean and covariance of features ⓘ |
| compares |
feature distributions of generated images
ⓘ
feature distributions of real images ⓘ |
| computedFrom |
covariance of feature embeddings
ⓘ
mean of feature embeddings ⓘ |
| criticizedFor |
bias from ImageNet-trained features
ⓘ
domain dependence ⓘ sensitivity to implementation details ⓘ |
| dependsOn |
Inception-v3 training data
ⓘ
choice of feature extractor ⓘ |
| describedIn |
Generative Adversarial Networks
ⓘ
surface form:
GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
|
| domain | image data ⓘ |
| featureLayer | Inception-v3 pool3 layer ⓘ |
| improvesUpon | Inception Score ⓘ |
| introducedBy |
Bernhard Nessler
ⓘ
Hubert Ramsauer ⓘ Martin Heusel ⓘ Sepp Hochreiter ⓘ Thomas Unterthiner ⓘ |
| introducedInField |
computer vision
ⓘ
machine learning ⓘ |
| lowerIsBetter | true ⓘ |
| measures |
distance between feature distributions
ⓘ
similarity between real and generated images ⓘ |
| metricType | full-reference metric ⓘ |
| output | non-negative real value ⓘ |
| publicationYear | 2017 ⓘ |
| relatedTo | Inception Score ⓘ |
| requires |
set of generated images
ⓘ
set of real images ⓘ |
| usedFor |
benchmarking generative models
ⓘ
comparing GAN architectures ⓘ evaluating sample diversity ⓘ evaluating sample quality ⓘ |
| usedIn |
evaluation of conditional GANs
ⓘ
research on diffusion models ⓘ research on image synthesis ⓘ |
| uses |
Inception architecture
ⓘ
surface form:
Inception network
pretrained Inception-v3 model ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: Fréchet Inception Distance Description of subject: Fréchet Inception Distance is a widely used quantitative metric that measures the similarity between real and generated images by comparing their feature distributions extracted from a pretrained Inception network.
Referenced by (5)
Full triples — surface form annotated when it differs from this entity's canonical label.
this entity surface form:
Fréchet Inception Score