# Coupling Perception and Reasoning in Federated Multimodal Graph Foundation Models

arXiv: https://arxiv.org/abs/2610.00277 · HTML: https://mlsanity.com/paper/2610.00277/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 24 references that are in the map.
- First posted on arXiv 24 Sep 2026.

## Builds on (most cited first, 22 of 22)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Communication-Efficient Learning of Deep Networks from Decentralized Data](https://mlsanity.com/paper/1602.05629/) (2016)
- [Shampoo: Preconditioned Stochastic Tensor Optimization](https://mlsanity.com/paper/1802.09568/) (2018)
- [Subgraph Federated Learning with Missing Neighbor Generation](https://mlsanity.com/paper/2106.13430/) (2021)
- [Federated Graph Classification over Non-IID Graphs](https://mlsanity.com/paper/2106.13423/) (2021)
- [Personalized Subgraph Federated Learning](https://mlsanity.com/paper/2206.10206/) (2022)
- [UniGraph2: Learning a Unified Embedding Space to Bind Multimodal Graphs](https://mlsanity.com/paper/2502.00806/) (2025)
- [FederatedScope-GNN: Towards a Unified, Comprehensive and Efficient Package for Federated Graph Learning](https://mlsanity.com/paper/2204.05562/) (2022)
- [NineRec: A Benchmark Dataset Suite for Evaluating Transferable Recommendation](https://mlsanity.com/paper/2309.07705/) (2023)
- [GRAPHGPT-O: Synergistic Multimodal Comprehension and Generation on Graphs](https://mlsanity.com/paper/2502.11925/) (2025)
- [Graph4MM: Weaving Multimodal Learning with Structural Information](https://mlsanity.com/paper/2510.16990/) (2025)
- [Towards Effective Federated Graph Foundation Model via Mitigating Knowledge Entanglement](https://mlsanity.com/paper/2505.12684/) (2025)
- [MM-OpenFGL: A Comprehensive Benchmark for Multimodal Federated Graph Learning](https://mlsanity.com/paper/2601.22416/) (2026)
- [FedBook: A Unified Federated Graph Foundation Codebook with Intra-domain and Inter-domain Knowledge Modeling](https://mlsanity.com/paper/2510.07755/) (2025)
- [Toward Effective Multimodal Graph Foundation Model: A Divide-and-Conquer Based Approach](https://mlsanity.com/paper/2602.04116/) (2026)
- [STAGE: Tackling Semantic Drift in Multimodal Federated Graph Learning](https://mlsanity.com/paper/2605.11919/) (2026)
- [PRISM: Topology-Aware Cross-Modal Imputation for Modality-Deficient Federated Graph Learning](https://mlsanity.com/paper/2606.09301/) (2026)
- [Rethinking Federated Graph Foundation Models: A Graph-Language Alignment-based Approach](https://mlsanity.com/paper/2601.21369/) (2026)
- [Towards Robust Federated Multimodal Graph Learning under Modality Heterogeneity](https://mlsanity.com/paper/2605.12584/) (2026)
- [FedLAB: Traceable Semantic Codebooks for Federated Multimodal Graph Foundation Learning](https://mlsanity.com/paper/2606.32016/) (2026)
- [Toward Federated Multimodal Graph Foundation Models: A Topology-Aware Multimodal Alignment Framework](https://mlsanity.com/paper/2607.15687/) (2026)
- [Towards Effective Federated Multimodal Graph Learning via Navigating Multifaceted Heterogeneity](https://mlsanity.com/paper/2608.00623/) (2026)

## Abstract (arXiv)

Federated multimodal graph foundation models (GFMs) aim to adapt pretrained multimodal models to decentralized graph data, where each client owns a private multimodal graph and cannot share raw information. These models typically combine a multimodal Encoder that extracts semantic evidence from heterogeneous modalities and a graph neural network (GNN) that performs relational reasoning over graph structures. However, existing federated GFM adaptation methods mainly update graph-side modules while keeping the multimodal Encoder frozen, limiting adaptation to how information is propagated while fixing what information is extracted. Through empirical studies, we reveal that Encoder and GNN adaptations are not independent: Encoder adaptation is affected by graph relations, while cross-client module swapping reveals substantial pairing sensitivity between separately parameterized Encoder and GNN updates. Motivated by this observation, we propose FedCORE, a federated adaptation framework that represents Encoder and GNN updates through a shared low-dimensional latent state. FedCORE jointly optimizes this core from multimodal and structural signals and performs federated evolution directly in the shared state space, preserving compatibility between perception and reasoning adaptations. Extensive experiments demonstrate that FedCORE reduces the Encoder--GNN pairing gap from 30.6 to 5.9, corresponding to an 80.7% reduction over independent joint adaptation.