CycleGAN
E290871
deep learning model
generative adversarial network architecture
image-to-image translation model
unpaired image-to-image translation method
CycleGAN is a type of generative adversarial network designed for unpaired image-to-image translation, enabling conversion between visual domains without requiring matched training examples.
All labels observed (4)
How this entity was disambiguated
This entity first appeared as the object of triple T2703880 — 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: CycleGAN Context triple: [Generative Adversarial Networks, notableVariant, CycleGAN]
-
A.
PixelRNN
PixelRNN is a deep generative model that uses recurrent neural networks to sequentially model and generate images pixel by pixel.
-
B.
PixelCNN
PixelCNN is a deep generative model that uses convolutional neural networks with autoregressive masking to model and generate images pixel by pixel.
-
C.
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.
-
D.
CLIP
CLIP is an OpenAI model that learns joint representations of images and text, enabling tasks like zero-shot image classification and natural language-based image retrieval.
-
E.
DALL·E
DALL·E is an AI model developed by OpenAI that generates images from natural language descriptions, enabling text-to-image synthesis.
- 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: CycleGAN Target entity description: CycleGAN is a type of generative adversarial network designed for unpaired image-to-image translation, enabling conversion between visual domains without requiring matched training examples.
-
A.
PixelRNN
PixelRNN is a deep generative model that uses recurrent neural networks to sequentially model and generate images pixel by pixel.
-
B.
PixelCNN
PixelCNN is a deep generative model that uses convolutional neural networks with autoregressive masking to model and generate images pixel by pixel.
-
C.
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.
-
D.
CLIP
CLIP is an OpenAI model that learns joint representations of images and text, enabling tasks like zero-shot image classification and natural language-based image retrieval.
-
E.
DALL·E
DALL·E is an AI model developed by OpenAI that generates images from natural language descriptions, enabling text-to-image synthesis.
- F. None of above. chosen
Statements (50)
| Predicate | Object |
|---|---|
| instanceOf |
deep learning model
ⓘ
generative adversarial network architecture ⓘ image-to-image translation model ⓘ unpaired image-to-image translation method ⓘ |
| basedOn | Generative Adversarial Networks ⓘ |
| canTranslateBetween |
apple images and orange images
ⓘ
horse images and zebra images ⓘ photo images and Monet-style paintings ⓘ summer photos and winter photos ⓘ |
| commonlyTrainedWith | Adam optimizer ⓘ |
| differsFrom | Pix2Pix by not requiring paired data ⓘ |
| doesNotRequire | paired training images ⓘ |
| evaluationDataset |
apple2orange dataset
ⓘ
horse2zebra dataset ⓘ photo2monet dataset ⓘ summer2winter_yosemite dataset ⓘ |
| field |
computer vision
ⓘ
generative modeling ⓘ machine learning ⓘ |
| hasComponent |
backward generator F:Y→X
ⓘ
discriminator DX for domain X ⓘ discriminator DY for domain Y ⓘ discriminator network ⓘ forward generator G:X→Y ⓘ generator network ⓘ |
| hasFullName |
Cycle-Consistent Generative Adversarial Network
ⓘ
linked to:
CycleGAN
|
| hasOpenSourceImplementation | official PyTorch implementation by authors ⓘ |
| implementedIn |
PyTorch
ⓘ
TensorFlow ⓘ |
| influenced | many subsequent unpaired translation methods ⓘ |
| introducedBy |
Alexei A. Efros
ⓘ
linked to:
Alexei Efros
Jun-Yan Zhu ⓘ Phillip Isola ⓘ Taesung Park ⓘ |
| introducedInPaper |
Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
ⓘ
linked to:
CycleGAN
|
| keyIdea |
enforce cycle consistency between forward and backward mappings
ⓘ
learn mappings between two visual domains without paired training data ⓘ |
| objective | unpaired image-to-image translation ⓘ |
| optimizationMethod | stochastic gradient descent variants ⓘ |
| publicationYear | 2017 ⓘ |
| publishedAtConference | ICCV 2017 ⓘ |
| relatedTo | Pix2Pix ⓘ |
| trainingDataRequirement | unpaired images from source and target domains ⓘ |
| uses |
adversarial loss
ⓘ
convolutional neural networks ⓘ cycle-consistency loss ⓘ identity loss ⓘ residual blocks ⓘ |
| usesArchitecture |
PatchGAN discriminator
ⓘ
ResNet-based generator ⓘ |
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: CycleGAN Description of subject: CycleGAN is a type of generative adversarial network designed for unpaired image-to-image translation, enabling conversion between visual domains without requiring matched training examples.
Referenced by (4)
Full triples — surface form annotated when it differs from this entity's canonical label.
linked to: CycleGAN
CycleGAN
→
introducedInPaper
→
Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
ⓘ
linked to: CycleGAN
linked to: CycleGAN