Contextualizing Meta-Learning via Learning to Decompose

TL;DR

This paper introduces LeadNet, a novel meta-learning approach that decomposes instance comparisons into multiple embedding spaces to better handle ambiguous similarities and diverse attributes in support sets.

Contribution

LeadNet uniquely decomposes comparison relationships into multiple embedding spaces, enabling context-aware meta-learning for improved handling of ambiguous and multi-attribute instances.

Findings

LeadNet outperforms existing methods in few-shot image classification.

It effectively manages ambiguous similarities in support sets.

Demonstrates superior performance in out-of-distribution recognition.

Abstract

Meta-learning has emerged as an efficient approach for constructing target models based on support sets. For example, the meta-learned embeddings enable the construction of target nearest-neighbor classifiers for specific tasks by pulling instances closer to their same-class neighbors. However, a single instance can be annotated from various latent attributes, making visually similar instances inside or across support sets have different labels and diverse relationships with others. Consequently, a uniform meta-learned strategy for inferring the target model from the support set fails to capture the instance-wise ambiguous similarity. To this end, we propose Learning to Decompose Network (LeadNet) to contextualize the meta-learned ``support-to-target'' strategy, leveraging the context of instances with one or mixed latent attributes in a support set. In particular, the comparison relationship between instances is decomposed w.r.t. multiple embedding spaces. LeadNet learns to automatically select the strategy associated with the right attribute via incorporating the change of comparison across contexts} with polysemous embeddings. We demonstrate the superiority of LeadNet in various applications, including exploring multiple views of confusing data, out-of-distribution recognition, and few-shot image classification.