ARC: Leveraging Compositional Representations for Cross-Problem Learning on VRPs

2025-12-21

AI Summary

Vehicle Routing Problems (VRPs) suffer from poor generalization across diverse real-world constraints and limited transferability across variants. Method: We propose the first analogy-consistency-driven attribute-disentangled representation framework for VRP. It employs a dual-stream attribute encoder to decompose problem representations into invariant semantic embeddings (IAE) and context-aware interaction effects (CIE), jointly optimized via contrastive learning and analogy-consistency regularization in a geometric embedding space to enable compositional representation learning. Contribution/Results: The framework supports zero-shot transfer and few-shot adaptation to unseen VRP variants without fine-tuning. It achieves state-of-the-art performance on in-distribution, zero-shot, few-shot, and real-world logistics benchmarks, significantly improving generalization and efficiency across VRP variants with heterogeneous constraints and structural configurations.

Technology Category

• Computer Vision: Representation Learning for Vision
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Machine Learning: Representation Learning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Vehicle Routing Problems (VRPs) with diverse real-world attributes have driven recent interest in cross-problem learning approaches that efficiently generalize across problem variants. We propose ARC (Attribute Representation via Compositional Learning), a cross-problem learning framework that learns disentangled attribute representations by decomposing them into two complementary components: an Intrinsic Attribute Embedding (IAE) for invariant attribute semantics and a Contextual Interaction Embedding (CIE) for attribute-combination effects. This disentanglement is achieved by enforcing analogical consistency in the embedding space to ensure the semantic transformation of adding an attribute (e.g., a length constraint) remains invariant across different problem contexts. This enables our model to reuse invariant semantics across trained variants and construct representations for unseen combinations. ARC achieves state-of-the-art performance across in-distribution, zero-shot generalization, few-shot adaptation, and real-world benchmarks.

Problem

• Learning disentangled attribute representations for diverse Vehicle Routing Problems
• Generalizing across VRP variants via invariant attribute semantics and contextual interactions
• Enabling zero-shot generalization to unseen attribute combinations in routing problems

Innovation

• Disentangled attribute representations via compositional learning
• Intrinsic and contextual embeddings for invariant semantics
• Analogical consistency for cross-problem generalization and adaptation