MC-INR: Efficient Encoding of Multivariate Scientific Simulation Data using Meta-Learning and Clustered Implicit Neural Representations

2025-07-03

AI Summary

Existing implicit neural representation (INR) methods exhibit three critical limitations in modeling multivariate scientific simulation data: rigid structural expressivity, univariate bias, and reliance on structured grids—leading to degraded performance on complex, real-world datasets. To address these issues, we propose MetaCluster-INR, a novel framework integrating meta-learning with hierarchical clustering. It introduces a residual-driven dynamic reclustering mechanism and a lightweight branched network architecture, enabling adaptive, fine-grained encoding of unstructured, multivariate, and highly heterogeneous scientific data. Our method eliminates grid constraints, supports coupled inter-variable modeling, and preserves local structural awareness. Extensive experiments across multiple scientific simulation datasets demonstrate that MetaCluster-INR significantly outperforms state-of-the-art INR baselines in reconstruction accuracy (PSNR improvement of 2.1–4.3 dB), memory efficiency (37%–62% reduction), and cross-dataset generalization.

Technology Category

• Machine Learning: Clustering
• Search and Optimization: Metareasoning and Metaheuristics
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Web data generation and simulation
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Implicit Neural Representations (INRs) are widely used to encode data as continuous functions, enabling the visualization of large-scale multivariate scientific simulation data with reduced memory usage. However, existing INR-based methods face three main limitations: (1) inflexible representation of complex structures, (2) primarily focusing on single-variable data, and (3) dependence on structured grids. Thus, their performance degrades when applied to complex real-world datasets. To address these limitations, we propose a novel neural network-based framework, MC-INR, which handles multivariate data on unstructured grids. It combines meta-learning and clustering to enable flexible encoding of complex structures. To further improve performance, we introduce a residual-based dynamic re-clustering mechanism that adaptively partitions clusters based on local error. We also propose a branched layer to leverage multivariate data through independent branches simultaneously. Experimental results demonstrate that MC-INR outperforms existing methods on scientific data encoding tasks.

Problem

• Handling multivariate data on unstructured grids
• Improving encoding of complex structures with meta-learning
• Enhancing performance via dynamic re-clustering mechanism

Innovation

• Meta-learning and clustered implicit neural representations
• Residual-based dynamic re-clustering mechanism
• Branched layer for multivariate data handling