LGM-Net: Learning to Generate Matching Networks for Few-Shot Learning

TL;DR

LGM-Net is a meta-learning framework that generates network parameters for few-shot classification tasks, enabling rapid adaptation to new tasks without additional tuning, demonstrated on Omniglot and miniImageNet datasets.

Contribution

This paper introduces LGM-Net, a novel meta-learning approach with TargetNet and MetaNet modules, and an intertask normalization strategy for effective few-shot learning.

Findings

Achieves competitive results on Omniglot and miniImageNet.

Effectively learns transferable prior knowledge across tasks.

Enables fast adaptation without further tuning.

Abstract

In this work, we propose a novel meta-learning approach for few-shot classification, which learns transferable prior knowledge across tasks and directly produces network parameters for similar unseen tasks with training samples. Our approach, called LGM-Net, includes two key modules, namely, TargetNet and MetaNet. The TargetNet module is a neural network for solving a specific task and the MetaNet module aims at learning to generate functional weights for TargetNet by observing training samples. We also present an intertask normalization strategy for the training process to leverage common information shared across different tasks. The experimental results on Omniglot and miniImageNet datasets demonstrate that LGM-Net can effectively adapt to similar unseen tasks and achieve competitive performance, and the results on synthetic datasets show that transferable prior knowledge is learned by the MetaNet module via mapping training data to functional weights. LGM-Net enables fast learning and adaptation since no further tuning steps are required compared to other meta-learning approaches.