Rebalanced Zero-shot Learning

TL;DR

This paper introduces a re-weighted loss function called ReMSE for zero-shot learning, addressing semantic prediction imbalance by modeling ZSL as an imbalanced regression problem, leading to improved performance.

Contribution

It formalizes imbalanced zero-shot learning as a regression problem and proposes ReMSE, a theoretically grounded re-weighted loss that balances semantic predictions across classes.

Findings

ReMSE effectively reduces semantic prediction imbalance.

The method outperforms existing state-of-the-art ZSL approaches.

Theoretical analysis confirms ReMSE's robustness.

Abstract

Zero-shot learning (ZSL) aims to identify unseen classes with zero samples during training. Broadly speaking, present ZSL methods usually adopt class-level semantic labels and compare them with instance-level semantic predictions to infer unseen classes. However, we find that such existing models mostly produce imbalanced semantic predictions, i.e. these models could perform precisely for some semantics, but may not for others. To address the drawback, we aim to introduce an imbalanced learning framework into ZSL. However, we find that imbalanced ZSL has two unique challenges: (1) Its imbalanced predictions are highly correlated with the value of semantic labels rather than the number of samples as typically considered in the traditional imbalanced learning; (2) Different semantics follow quite different error distributions between classes. To mitigate these issues, we first formalize ZSL as an imbalanced regression problem which offers empirical evidences to interpret how semantic labels lead to imbalanced semantic predictions. We then propose a re-weighted loss termed Re-balanced Mean-Squared Error (ReMSE), which tracks the mean and variance of error distributions, thus ensuring rebalanced learning across classes. As a major contribution, we conduct a series of analyses showing that ReMSE is theoretically well established. Extensive experiments demonstrate that the proposed method effectively alleviates the imbalance in semantic prediction and outperforms many state-of-the-art ZSL methods. Our code is available at https://github.com/FouriYe/ReZSL-TIP23.