Few-Shot Learning

The general framework for few-shot learning via kernel HyperNetworks—the fusion of kernels and hypernetwork paradigm is proposed, and it is shown that incorporating Bayesian neural networks into the general framework, an approach called BayesHyperShot, solves this issue.

The proposed HyperShot - the fusion of kernels and hypernetwork paradigm aims to switch the classification module parameters depending on the task’s embedding, and relies on relations between embeddings of the support examples instead of direct feature values provided by the backbone models.