Prototypical Networks
E899064
few-shot learning method
metric-based meta-learning method
neural network based model
supervised learning algorithm
Prototypical Networks are a few-shot learning method that represents each class by the mean of its embedded support examples and classifies queries based on distances to these learned prototypes in embedding space.
All labels observed (1)
| Label | Occurrences |
|---|---|
| Prototypical Networks canonical | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T11003617 — 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: Prototypical Networks Context triple: [Matching Networks for One Shot Learning, influenced, Prototypical Networks]
-
A.
Matching Networks for One Shot Learning
"Matching Networks for One Shot Learning" is a seminal deep learning paper that introduced a metric-based approach for one-shot image classification using attention and memory-augmented neural networks.
-
B.
Pointer Networks
Pointer Networks are a type of neural network architecture that uses attention mechanisms to output discrete positions in an input sequence, enabling solutions to combinatorial problems like sorting and the traveling salesman problem.
-
C.
Learning Transferable Architectures for Scalable Image Recognition
"Learning Transferable Architectures for Scalable Image Recognition" is a research paper that introduced NASNet, a neural architecture search–designed convolutional network that achieved state-of-the-art performance on large-scale image recognition tasks.
-
D.
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.
-
E.
Neural Architecture Search
Neural Architecture Search is an automated machine learning technique that uses algorithms to design and optimize neural network architectures without extensive human intervention.
- 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: Prototypical Networks Target entity description: Prototypical Networks are a few-shot learning method that represents each class by the mean of its embedded support examples and classifies queries based on distances to these learned prototypes in embedding space.
-
A.
Matching Networks for One Shot Learning
"Matching Networks for One Shot Learning" is a seminal deep learning paper that introduced a metric-based approach for one-shot image classification using attention and memory-augmented neural networks.
-
B.
Pointer Networks
Pointer Networks are a type of neural network architecture that uses attention mechanisms to output discrete positions in an input sequence, enabling solutions to combinatorial problems like sorting and the traveling salesman problem.
-
C.
Learning Transferable Architectures for Scalable Image Recognition
"Learning Transferable Architectures for Scalable Image Recognition" is a research paper that introduced NASNet, a neural architecture search–designed convolutional network that achieved state-of-the-art performance on large-scale image recognition tasks.
-
D.
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.
-
E.
Neural Architecture Search
Neural Architecture Search is an automated machine learning technique that uses algorithms to design and optimize neural network architectures without extensive human intervention.
- F. None of above. chosen
Statements (47)
| Predicate | Object |
|---|---|
| instanceOf |
few-shot learning method
ⓘ
metric-based meta-learning method ⓘ neural network based model ⓘ supervised learning algorithm ⓘ |
| advantage |
non-parametric classification layer
ⓘ
simple and efficient inference ⓘ |
| appliedTo |
Omniglot dataset
ⓘ
image classification ⓘ miniImageNet dataset ⓘ |
| assumes | embedding space where classes form tight clusters ⓘ |
| basedOn |
metric learning
ⓘ
nearest prototype classification ⓘ |
| canUse |
cosine distance
ⓘ
squared Euclidean distance ⓘ |
| citationCountCategory | highly cited in few-shot learning literature ⓘ |
| classifiesBy | distance to class prototypes ⓘ |
| codeAvailableAs |
open-source implementations in PyTorch
ⓘ
open-source implementations in TensorFlow ⓘ |
| comparedWith |
MAML
NERFINISHED
ⓘ
Matching Networks NERFINISHED ⓘ |
| computes | prototype for each class in embedding space ⓘ |
| describedInPaper | Prototypical Networks for Few-shot Learning NERFINISHED ⓘ |
| evaluationProtocol | episodic evaluation matching training setup ⓘ |
| extendedTo |
cross-domain few-shot learning
ⓘ
semi-supervised few-shot learning variants ⓘ zero-shot learning variants ⓘ |
| inferenceStep |
apply softmax over negative distances
ⓘ
compute distances from query embeddings to prototypes ⓘ compute prototype for each class from support set ⓘ |
| inspired | many subsequent prototype-based few-shot methods ⓘ |
| learningObjective | learn embedding where examples cluster around class prototypes ⓘ |
| optimizedWith | cross-entropy loss over episodic tasks ⓘ |
| proposedBy |
Jake Snell
NERFINISHED
ⓘ
Kevin Swersky NERFINISHED ⓘ Richard Zemel NERFINISHED ⓘ |
| publicationYear | 2017 ⓘ |
| publishedIn | Neural Information Processing Systems (NeurIPS) NERFINISHED ⓘ |
| representsClassAs | mean of embedded support examples ⓘ |
| supports | N-way K-shot classification ⓘ |
| trainingRegime | episodic few-shot tasks ⓘ |
| typicallyUses | Euclidean distance ⓘ |
| uses |
class prototypes
ⓘ
distance metric ⓘ embedding function ⓘ episodic training ⓘ query set ⓘ support set ⓘ |
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: Prototypical Networks Description of subject: Prototypical Networks are a few-shot learning method that represents each class by the mean of its embedded support examples and classifies queries based on distances to these learned prototypes in embedding space.
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.