SEED: Targeted Data Selection by Weighted Independent Set

2026-05-15

AI Summary

This work addresses the challenge of efficiently selecting high-quality, diverse, and compact data subsets from large-scale training corpora. The authors formulate data selection as a weighted independent set problem on a similarity graph, jointly optimizing sample quality and diversity through graph-structured modeling. They introduce a novel node value calibration mechanism to emphasize task-relevant signals and propose a local-scale normalization strategy to mitigate structural imbalances in the graph caused by cross-domain distribution shifts. Their approach integrates bilateral salient subspace influence estimation with adaptive edge thresholding based on local density. Evaluated across instruction tuning, vision instruction tuning, and semantic segmentation tasks, the method consistently outperforms state-of-the-art baselines and enables the construction of Honeybee-Remake-SEED-200K, a high-quality multimodal dataset.

Abstract

Data selection seeks to identify a compact yet informative subset from large-scale training corpora, balancing sample quality against collection diversity. We formulate this problem as a Weighted Independent Set (WIS) on a similarity graph, where nodes represent data samples weighted by influence, and edges connect semantically redundant pairs. This formulation naturally yields subsets that are simultaneously high-quality and diverse. However, two challenges arise in practice: naive node weights fail to distinguish informative signals from gradient noise, and edge construction under heterogeneous domain distributions produces structurally imbalanced graphs that bias selection toward sparse regions. To address these issues, we introduce two principled refinements from a unified graph perspective: (1) \textit{node value calibration} that restricts influence estimation to the bilateral salient subspace to ground node importance in task-relevant signals rather than surface-level statistics; (2) \textit{local scale normalization} that adapts edge thresholds to local neighborhood density, mitigating graph imbalance induced by cross-domain distribution shifts. Together, these components yield a robust and scalable data selection pipeline dubbed SEED. We further construct \texttt{Honeybee-Remake-SEED-200K}, a compact multimodal dataset curated by SEED. Extensive experiments show that SEED consistently outperforms state-of-the-art methods on instruction tuning, visual instruction tuning, and semantic segmentation across diverse model families.

Problem

• data selection
• Weighted Independent Set
• similarity graph
• domain distribution shift
• sample diversity

Innovation

• Weighted Independent Set
• node value calibration
• local scale normalization
• data selection
• graph-based optimization