Adaptive manifold for imbalanced transductive few-shot learning

TL;DR

This paper introduces Adaptive Manifold, a novel algorithm for imbalanced transductive few-shot learning that leverages manifold similarity to improve class prediction, outperforming existing methods on multiple benchmarks.

Contribution

The paper proposes a new manifold-based approach with a tunable loss function for imbalanced transductive few-shot learning, showing significant improvements over Euclidean distance and prior methods.

Findings

Manifold similarity outperforms Euclidean distance, especially in 1-shot learning.

The method achieves state-of-the-art results on miniImageNet, tieredImageNet, and CUB datasets.

In some cases, it outperforms previous best methods by up to 4.2%.

Abstract

Transductive few-shot learning algorithms have showed substantially superior performance over their inductive counterparts by leveraging the unlabeled queries. However, the vast majority of such methods are evaluated on perfectly class-balanced benchmarks. It has been shown that they undergo remarkable drop in performance under a more realistic, imbalanced setting. To this end, we propose a novel algorithm to address imbalanced transductive few-shot learning, named Adaptive Manifold. Our method exploits the underlying manifold of the labeled support examples and unlabeled queries by using manifold similarity to predict the class probability distribution per query. It is parameterized by one centroid per class as well as a set of graph-specific parameters that determine the manifold. All parameters are optimized through a loss function that can be tuned towards class-balanced or imbalanced distributions. The manifold similarity shows substantial improvement over Euclidean distance, especially in the 1-shot setting. Our algorithm outperforms or is on par with other state of the art methods in three benchmark datasets, namely miniImageNet, tieredImageNet and CUB, and three different backbones, namely ResNet-18, WideResNet-28-10 and DenseNet-121. In certain cases, our algorithm outperforms the previous state of the art by as much as 4.2%.