matching networks

E899063

few-shot learning method metric-based meta-learning method neural network architecture

Matching networks are a neural network architecture designed to perform one-shot learning by leveraging metric-based comparisons between support and query examples.

Try in SPARQL Jump to: Surface forms Statements Referenced by

Observed surface forms (1)

Surface form	Occurrences
Matching Networks	0

Statements (39)

Predicate	Object
instanceOf	few-shot learning method ⓘ metric-based meta-learning method ⓘ neural network architecture ⓘ
advantage	good performance with very few labeled examples ⓘ
appliedTo	Omniglot dataset ⓘ image classification ⓘ miniImageNet dataset NERFINISHED ⓘ
category	non-parametric prediction over support set ⓘ
citationYear	2016 ⓘ
coreIdea	classify queries by comparing them to labeled support examples in an embedding space ⓘ
designedFor	few-shot classification ⓘ one-shot learning ⓘ
developedAt	DeepMind NERFINISHED ⓘ
evaluationProtocol	N-way K-shot classification episodes ⓘ
handles	variable-sized support sets ⓘ
inputIncludes	query set ⓘ support set ⓘ
inspired	subsequent metric-based few-shot methods ⓘ
introducedInPaper	Matching Networks for One Shot Learning NERFINISHED ⓘ
keyComponent	attention-based classifier ⓘ embedding network for query examples ⓘ embedding network for support examples ⓘ
learningType	supervised learning ⓘ
optimization	trained end-to-end with gradient descent ⓘ
outputs	label distribution over support set labels ⓘ
proposedBy	Charles Blundell NERFINISHED ⓘ Daan Wierstra NERFINISHED ⓘ Oriol Vinyals NERFINISHED ⓘ Timothy Lillicrap NERFINISHED ⓘ
publishedAtConference	NeurIPS 2016 NERFINISHED ⓘ
relatedTo	Prototypical Networks NERFINISHED ⓘ Siamese networks NERFINISHED ⓘ meta-learning ⓘ
trainingParadigm	episodic training ⓘ
uses	attention kernel over support embeddings ⓘ attention mechanism ⓘ cosine similarity ⓘ embedding functions ⓘ metric-based comparisons ⓘ

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Matching Networks for One Shot Learning → introducesConcept → matching networks ⓘ