This work introduces ICE (Interaction-aware Clifford Encoder), a multimodal graph foundation model built on a node-indexed Clifford latent field that ranks first in all 30 reported supervised and few-shot comparisons.
Abstract
Multimodal attributed graphs connect entities, visual content, language, and observed relations. Learning one foundation across such graphs requires more than compressing each node into a fused Euclidean vector. The representation must preserve entity semantics, construct interaction state from graph neighborhoods, and expose that state to prediction units with different geometry. Our empirical study shows why these requirements are inseparable. Higher-grade channels recover pair relations across the foundation graphs, specialized queries reveal information hidden by a generic readout, and rigid blade isolation removes cross-grade capacity. We therefore introduce ICE (Interaction-aware Clifford Encoder), a multimodal graph foundation model built on a node-indexed Clifford latent field. Topology, text, and images enter explicit Cl(3) addresses. Edge-aware geometric products transform these directions into scalar, bivector, and trivector relations over observed neighborhoods. A protected Grade-1 route preserves entity semantics, while the full grade and depth bank remains available to fresh node and link heads. We establish exact cross-grade reachability, node-permutation equivariance, and a bound on the task residual around the semantic score. Experiments span one shared foundation over eleven graphs, six node-classification datasets, three link-prediction datasets, and matched few-shot tasks. ICE ranks first in all 30 reported supervised and few-shot comparisons. Core removals reduce every task summary, and mechanism controls connect the gains to higher-order transport, retained multidepth structure, semantic protection, and direct field access.
While foundation models have revolutionized natural language processing and computer vision by leveraging universal vocabularies, Graph Machine Learning (GML) remains fractured due to the absence of a unified feature and structural representation across diverse domains. Existing works claiming to be Graph Foundation Mo...
Ben Finkelshtein, André Linhares, Petar Veličković et al.· arXiv.org· 1 citation
A dual-channel knowledge fusion framework, named MoKGNN, to simultaneously leverage the generic semantic knowledge from LLMs and the task adaptation capability of TLM, and a Knowledge Alignment Network to align the LLM embeddings with the TLM representations.
Hao Yan, Chao-Zhuo Li, Jun Yin et al.· ACM Transactions on Intellig...· 0 citations
SCGFM-ART, a structure-centric GFM framework that aligns arbitrary graphs onto a shared relational atlas via Amortized Relational Transport, achieves state-of-the-art transferability and establishes coordinate fidelity bounds, prove stability under predicted transport plans, and derive an amortized coverage bound that...
Multimodal graph foundation models (MGFMs) seek to learn generalizable representations from large-scale graphs with heterogeneous node modalities. However, real-world Multimodal-Attributed Graphs (MAGs) often contain incomplete node attributes, limiting the scale and diversity of available pretraining corpora. Besides,...
Xun-Kai Li, Chen-Xi Wan, Yin-Lin Zhu et al.· 0 citations
Lightweight sentence embedding models are critical for large-scale retrieval and semantic matching, yet aggressive compression often disrupts the semantic geometry learned by powerful teachers. Existing distillation methods primarily rely on point-wise feature matching or task-specific supervision, which fails to prese...
Chen Peng, Xiang-Bo Kong, Xue Qiao et al.· International journal of pat...· 0 citations
Multimodal large language models process the complete visual token sequence, increasing prefill cost even when many tokens are irrelevant to the instruction. Token-wise saliency captures limited inter-token dependence; graph-based selection instead depends on how relations govern importance propagation and ranking. We...
Xiao-Yang Ma, Chi Wang, Hong Sun· 0 citations
Related blog posts
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026