Class Distribution Shifts in Zero-Shot Learning: Learning Robust Representations
This addresses robustness issues in zero-shot learning for real-world applications like person identification, but it is incremental as it builds on existing out-of-distribution methods.
The paper tackles the problem of class distribution shifts in zero-shot learning, where unseen classes during deployment differ from training distributions, and proposes an algorithm that improves generalization to diverse class distributions, as shown in simulations and real-world datasets.
Zero-shot learning methods typically assume that the new, unseen classes encountered during deployment come from the same distribution as the the classes in the training set. However, real-world scenarios often involve class distribution shifts (e.g., in age or gender for person identification), posing challenges for zero-shot classifiers that rely on learned representations from training classes. In this work, we propose and analyze a model that assumes that the attribute responsible for the shift is unknown in advance. We show that in this setting, standard training may lead to non-robust representations. To mitigate this, we develop an algorithm for learning robust representations in which (a) synthetic data environments are constructed via hierarchical sampling, and (b) environment balancing penalization, inspired by out-of-distribution problems, is applied. We show that our algorithm improves generalization to diverse class distributions in both simulations and experiments on real-world datasets.