Structure-Aware Prototype Guided Trusted Multi-View Classification

2025-11-25

AI Summary

This work addresses three core challenges in trustworthy multi-view classification (TMVC) under heterogeneous and conflicting multi-source information: inconsistent inter-view relationships, prohibitively high computational cost of global modeling, and lack of class-space consistency guarantees during evidence aggregation. To this end, we propose a structure-aware, prototype-guided TMVC framework. Our method replaces conventional global dense graph modeling with learnable prototypes that capture local neighborhood structures within each view. Cross-view structural alignment and dynamic evidence aggregation are performed explicitly in the prototype space, thereby enforcing intra-class consistency. Extensive experiments on multiple benchmark multi-view datasets demonstrate that our approach significantly outperforms state-of-the-art trustworthy multi-view methods in classification accuracy, robustness to conflicts and heterogeneity, and computational efficiency. The proposed framework establishes a novel paradigm for reliable decision-making under heterogeneous and conflicting multi-source scenarios.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Trustworthy multi-view classification (TMVC) addresses the challenge of achieving reliable decision-making in complex scenarios where multi-source information is heterogeneous, inconsistent, or even conflicting. Existing TMVC approaches predominantly rely on globally dense neighbor relationships to model intra-view dependencies, leading to high computational costs and an inability to directly ensure consistency across inter-view relationships. Furthermore, these methods typically aggregate evidence from different views through manually assigned weights, lacking guarantees that the learned multi-view neighbor structures are consistent within the class space, thus undermining the trustworthiness of classification outcomes. To overcome these limitations, we propose a novel TMVC framework that introduces prototypes to represent the neighbor structures of each view. By simplifying the learning of intra-view neighbor relations and enabling dynamic alignment of intra- and inter-view structure, our approach facilitates more efficient and consistent discovery of cross-view consensus. Extensive experiments on multiple public multi-view datasets demonstrate that our method achieves competitive downstream performance and robustness compared to prevalent TMVC methods.

Problem

• Achieving reliable classification with heterogeneous multi-source information
• Reducing computational costs from dense neighbor relationship modeling
• Ensuring consistency across inter-view relationships for trustworthy outcomes

Innovation

• Prototypes represent neighbor structures per view
• Dynamic alignment of intra- and inter-view structures
• Simplifies learning while ensuring cross-view consistency