SMA uses a submodular mutual information objective on data sets to deliver competitive zero-shot classification and retrieval performance on CLIP benchmarks with only tens of thousands of samples, orders of magnitude fewer than standard approaches.
Elsevier, 2005
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
A parameterized family of alpha-orderings unifies pendent-pair and flat-pair results for symmetric submodular minimization and supports an O(n^3) oracle-based contraction algorithm.
citing papers explorer
-
SMA: Submodular Modality Aligner For Data Efficient Multimodal Learning
SMA uses a submodular mutual information objective on data sets to deliver competitive zero-shot classification and retrieval performance on CLIP benchmarks with only tens of thousands of samples, orders of magnitude fewer than standard approaches.
-
A Unified Approach to Minimizing Symmetric Submodular Functions
A parameterized family of alpha-orderings unifies pendent-pair and flat-pair results for symmetric submodular minimization and supports an O(n^3) oracle-based contraction algorithm.