Transitivity Recovering Decompositions: Interpretable and Robust Fine-Grained Relationships

Published: 21 Sept 2023, Last Modified: 02 Nov 2023NeurIPS 2023 posterEveryoneRevisionsBibTeX
Keywords: Interpretability, Robustness, Fine-Grained Representation Learning, Graph Theory, Information Theory
TL;DR: We propose an approach for learning robust and interpretable equivalents of abstract multi-view relational representations by showing that such abstractions are nothing but a way of encoding the transitive relationships across views.
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
Supplementary Material: pdf
Submission Number: 2235