Multimodal Representation-disentangled Information Bottleneck for Multimodal Recommendation

2025-09-24

AI Summary

Multimodal recommendation faces dual challenges: interference from modality-specific noise and insufficient modeling of cross-modal couplings. To address these, this paper proposes a multimodal recommendation framework grounded in the information bottleneck principle and disentangled representation learning. It innovatively decomposes cross-modal information into three distinct components—uniqueness (modality-specific), redundancy (shared across modalities), and synergy (complementary interactions)—and introduces a triple-constraint objective: (i) modality-specific regularization to preserve unique semantics, (ii) redundancy minimization to suppress noisy overlap, and (iii) synergy-consistency modeling to enhance complementary fusion. Extensive experiments on three benchmark datasets demonstrate consistent improvements over state-of-the-art models, achieving an average 4.2% gain in Recall@20. The results validate the framework’s robustness and generalizability, establishing a novel paradigm for controllable, disentangled multimodal representation learning and fusion.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Multimodal data has significantly advanced recommendation systems by integrating diverse information sources to model user preferences and item characteristics. However, these systems often struggle with redundant and irrelevant information, which can degrade performance. Most existing methods either fuse multimodal information directly or use rigid architectural separation for disentanglement, failing to adequately filter noise and model the complex interplay between modalities. To address these challenges, we propose a novel framework, the Multimodal Representation-disentangled Information Bottleneck (MRdIB). Concretely, we first employ a Multimodal Information Bottleneck to compress the input representations, effectively filtering out task-irrelevant noise while preserving rich semantic information. Then, we decompose the information based on its relationship with the recommendation target into unique, redundant, and synergistic components. We achieve this decomposition with a series of constraints: a unique information learning objective to preserve modality-unique signals, a redundant information learning objective to minimize overlap, and a synergistic information learning objective to capture emergent information. By optimizing these objectives, MRdIB guides a model to learn more powerful and disentangled representations. Extensive experiments on several competitive models and three benchmark datasets demonstrate the effectiveness and versatility of our MRdIB in enhancing multimodal recommendation.

Problem

• Filtering redundant and irrelevant information in multimodal recommendation systems
• Modeling complex interplay between different information modalities effectively
• Learning disentangled representations to capture unique and synergistic information

Innovation

• Multimodal Information Bottleneck compresses representations to filter noise
• Decomposes information into unique, redundant, and synergistic components
• Uses specific learning objectives to guide disentangled representation learning