Transitivity Recovering Decompositions: Interpretable and Robust Fine-Grained Relationships

TL;DR

This paper introduces Transitivity Recovering Decompositions (TRD), a method that interprets and robustly recovers fine-grained relational structures in images as interpretable graphs, improving interpretability and robustness in representation learning.

Contribution

The paper proposes TRD, a novel graph-based algorithm that explicitly recovers transitive relationships in image representations, enhancing interpretability and robustness without post-hoc adjustments.

Findings

TRD effectively recovers interpretable relationships.

TRD is robust to noisy views.

TRD matches or surpasses state-of-the-art performance.

Abstract

Recent advances in fine-grained representation learning leverage local-to-global (emergent) relationships for achieving state-of-the-art results. The relational representations relied upon by such methods, however, are abstract. We aim to deconstruct this abstraction by expressing them as interpretable graphs over image views. We begin by theoretically showing that abstract relational representations are nothing but a way of recovering transitive relationships among local views. Based on this, we design Transitivity Recovering Decompositions (TRD), a graph-space search algorithm that identifies interpretable equivalents of abstract emergent relationships at both instance and class levels, and with no post-hoc computations. We additionally show that TRD is provably robust to noisy views, with empirical evidence also supporting this finding. The latter allows TRD to perform at par or even better than the state-of-the-art, while being fully interpretable. Implementation is available at https://github.com/abhrac/trd.