What Remains of Visual Semantic Embeddings
This work addresses the need for fair evaluation in ZSL tasks, which is important for researchers in computer vision and machine learning, but it is incremental as it focuses on benchmarking rather than proposing a new method.
The paper tackles the problem of evaluating visual semantic embedding models in zero-shot learning (ZSL) by introducing a new benchmark split of tiered-ImageNet to avoid structural flaws, and shows that current models struggle with encoding semantic relationships from word analogy and hierarchy.
Zero shot learning (ZSL) has seen a surge in interest over the decade for its tight links with the mechanism making young children recognize novel objects. Although different paradigms of visual semantic embedding models are designed to align visual features and distributed word representations, it is unclear to what extent current ZSL models encode semantic information from distributed word representations. In this work, we introduce the split of tiered-ImageNet to the ZSL task, in order to avoid the structural flaws in the standard ImageNet benchmark. We build a unified framework for ZSL with contrastive learning as pre-training, which guarantees no semantic information leakage and encourages linearly separable visual features. Our work makes it fair for evaluating visual semantic embedding models on a ZSL setting in which semantic inference is decisive. With this framework, we show that current ZSL models struggle with encoding semantic relationships from word analogy and word hierarchy. Our analyses provide motivation for exploring the role of context language representations in ZSL tasks.