Neighborhood Sensitive Mapping for Zero-Shot Classification using Independently Learned Semantic Embeddings

TL;DR

This paper introduces a two-step approach for zero-shot classification that learns a property embedding space to better capture label relationships and improve classifier generalization, addressing limitations of semantic embeddings.

Contribution

The paper proposes a novel two-step method that learns a property embedding space to enhance zero-shot classification by capturing data-dependent label relationships.

Findings

Improved zero-shot classification accuracy.

Effective reduction in generalization error.

Better capture of label relationships compared to semantic embeddings.

Abstract

In a traditional setting, classifiers are trained to approximate a target function $f:X \rightarrow Y$ where at least a sample for each $y \in Y$ is presented to the training algorithm. In a zero-shot setting we have a subset of the labels $\hat{Y} \subset Y$ for which we do not observe any corresponding training instance. Still, the function $f$ that we train must be able to correctly assign labels also on $\hat{Y}$. In practice, zero-shot problems are very important especially when the label set is large and the cost of editorially label samples for all possible values in the label set might be prohibitively high. Most recent approaches to zero-shot learning are based on finding and exploiting relationships between labels using semantic embeddings. We show in this paper that semantic embeddings, despite being very good at capturing relationships between labels, are not very good at capturing the relationships among labels in a data-dependent manner. For this reason, we propose a novel two-step process for learning a zero-shot classifier. In the first step, we learn what we call a \emph{property embedding space} capturing the "\emph{learnable}" features of the label set. Then, we exploit the learned properties in order to reduce the generalization error for a linear nearest neighbor-based classifier.