What question did this study set out to answer?

This research aims to enhance few-shot classification by using prototypical networks that generalize well to unseen classes.

March 15, 2017Open Access

Prototypical Networks for Few-shot Learning

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved few-shot classification accuracy in limited-data scenarios, suggesting effective new architecture approaches.

This research aims to enhance few-shot classification by using prototypical networks that generalize well to unseen classes.
Developed prototypical networks that learn a metric space for classification.
Compared with other approaches to analyze design decisions for effectiveness.
Extended prototypical networks to zero-shot learning applications.
Achieved state-of-the-art results on the CU-Birds dataset for zero-shot learning.
Demonstrated substantial improvements over complex architectures using simple design choices.

Snell et al. (2017) studied this question.

View Full Paper

Ask AI

Bookmark

Randomized trial demonstrates improved few-shot classification accuracy in limited-data scenarios, suggesting effective new architecture approaches.