Combat Data Shift in Few-shot Learning with Knowledge Graph

TL;DR

This paper introduces a novel metric-based meta-learning framework utilizing knowledge graphs to effectively address data shift issues in few-shot learning scenarios, improving performance on benchmarks and new datasets.

Contribution

It proposes a new framework that combines task-specific and task-shared representations with knowledge graphs to combat data shift in few-shot learning.

Findings

Achieves remarkable performance on benchmarks.

Effectively handles data shift within and between tasks.

Outperforms existing methods on new challenging datasets.

Abstract

Many few-shot learning approaches have been designed under the meta-learning framework, which learns from a variety of learning tasks and generalizes to new tasks. These meta-learning approaches achieve the expected performance in the scenario where all samples are drawn from the same distributions (i.i.d. observations). However, in real-world applications, few-shot learning paradigm often suffers from data shift, i.e., samples in different tasks, even in the same task, could be drawn from various data distributions. Most existing few-shot learning approaches are not designed with the consideration of data shift, and thus show downgraded performance when data distribution shifts. However, it is non-trivial to address the data shift problem in few-shot learning, due to the limited number of labeled samples in each task. Targeting at addressing this problem, we propose a novel metric-based meta-learning framework to extract task-specific representations and task-shared representations with the help of knowledge graph. The data shift within/between tasks can thus be combated by the combination of task-shared and task-specific representations. The proposed model is evaluated on popular benchmarks and two constructed new challenging datasets. The evaluation results demonstrate its remarkable performance.