aiqbench analysis
GraphSelect introduces a novel method for optimizing input selection in multimodal graph inference by dynamically exchanging text and image representations to minimize prediction loss. This approach significantly reduces computational overhead while maintaining high classification accuracy, demonstrating that retaining only 20% of candidate inputs causes negligible performance degradation compared to full-input models. The technique advances the efficiency of large-scale graph neural networks without relying on standard attribution methods.
GraphSelect for Budgeted Representation Selection in Multimodal Graph Inference
arXiv:2609.33561v1 Announce Type: new Abstract: Multimodal graph predictors combine text, images, and relations to classify connected entities. How much of this input is needed to preserve their predictions? We study budgeted representation selection, which chooses a subset of candidate text and image vectors under a separate capacity for each modality. Predictions from the complete candidate input define the classes to preserve. The challenge is that a representation's contribution depends on the other selected inputs, while graph propagation extends its effects across nodes. Our empirical study shows that candidate rankings change with the selected input, while predicted probabilities remain informative after the class stops changing. Updating scores improves selection, and exchanging inputs can improve a subset whose capacity is already filled. These findings lead to GraphSelect, which starts from individual candidate gains and refines the subset through jointly evaluated exchanges. It screens promising removals and additions, accepts an exchange when it reduces the prediction loss, and updates the scores. Experiments on six graphs show higher mean objective recovery than six attribution and explanation methods adapted to the selection task. Across nine trained architectures on two graphs, retaining 20% of the candidate representations per modality gives a mean accuracy drop of 0.10 percentage points relative to full candidate input, preserving classification performance with substantially fewer text and image representations.
Abstract:Multimodal graph predictors combine text, images, and relations to classify connected entities. How much of this input is needed to preserve their predictions? We study budgeted representation selection, which chooses a subset of candidate text and image vectors under a separate capacity for each modality. Predictions from the complete candidate input define the classes to preserve. The challenge is that a representation's contribution depends on the other selected inputs, while graph propagation extends its effects across nodes. Our empirical study shows that candidate rankings change with the selected input, while predicted probabilities remain informative after the class stops changing. Updating scores improves selection, and exchanging inputs can improve a subset whose capacity is already filled. These findings lead to GraphSelect, which starts from individual candidate gains and refines the subset through jointly evaluated exchanges. It screens promising removals and additions, accepts an exchange when it reduces the prediction loss, and updates the scores. Experiments on six graphs show higher mean objective recovery than six attribution and explanation methods adapted to the selection task. Across nine trained architectures on two graphs, retaining 20% of the candidate representations per modality gives a mean accuracy drop of 0.10 percentage points relative to full candidate input, preserving classification performance with substantially fewer text and image representations.
Machine Learning (cs.LG)
arXiv:2609.33561 [cs.LG]
(or arXiv:2609.33561v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.33561 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 306 words
Related coverage
Q-CueGraph: Query-Conditioned Visual Evidence Graphs for Multimodal Reasoning
arXiv · Aug 6, 2026
Links #5: 2026/07
LessWrong · Jul 31, 2026
ViSR-KGC: Visual Subgraph Reasoning with Vision-Language Models for Multimodal Knowledge Graph Completion
arXiv · Aug 7, 2026
MGDT: MLLM-Guided Diffusion Transformer with Relation-Adaptive Mixture-of-Experts for Multimodal Knowledge Graph Completion
arXiv · Jul 20, 2026