Physics-Guided Dual Implicit Neural Representations for Source Separation

2025-07-07

AI Summary

Advanced experiments—such as neutron scattering—often yield signals obscured by complex, unknown backgrounds and nonlinear distortions, with scarce or no labeled ground-truth data for supervised learning. Method: We propose Dual Implicit Neural Representations (DINR), a physics-guided self-supervised framework that jointly models signal and background in an end-to-end manner, without requiring predefined dictionaries or manual annotations. DINR employs a coupled dual-network architecture optimized via reconstruction loss and analytically tractable regularization parameters, enabling efficient multi-dimensional inverse modeling. Results: Evaluated on four-dimensional neutron scattering data, DINR robustly separates highly heterogeneous backgrounds from morphologically diverse physical signals, achieving substantial SNR improvement and enhanced feature interpretability. It establishes a new paradigm for experimental data decoupling under minimal prior assumptions—eliminating the need for domain-specific handcrafted features or explicit background parameterization.

Technology Category

• Machine Learning: Representation Learning
• Computer Vision: Representation Learning for Vision
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Machine learning and data science for the Web

Abstract

Significant challenges exist in efficient data analysis of most advanced experimental and observational techniques because the collected signals often include unwanted contributions--such as background and signal distortions--that can obscure the physically relevant information of interest. To address this, we have developed a self-supervised machine-learning approach for source separation using a dual implicit neural representation framework that jointly trains two neural networks: one for approximating distortions of the physical signal of interest and the other for learning the effective background contribution. Our method learns directly from the raw data by minimizing a reconstruction-based loss function without requiring labeled data or pre-defined dictionaries. We demonstrate the effectiveness of our framework by considering a challenging case study involving large-scale simulated as well as experimental momentum-energy-dependent inelastic neutron scattering data in a four-dimensional parameter space, characterized by heterogeneous background contributions and unknown distortions to the target signal. The method is found to successfully separate physically meaningful signals from a complex or structured background even when the signal characteristics vary across all four dimensions of the parameter space. An analytical approach that informs the choice of the regularization parameter is presented. Our method offers a versatile framework for addressing source separation problems across diverse domains, ranging from superimposed signals in astronomical measurements to structural features in biomedical image reconstructions.

Problem

• Separating physical signals from complex background distortions
• Self-supervised source separation without labeled data
• Handling four-dimensional parameter space with varying signal characteristics

Innovation

• Dual implicit neural networks for separation
• Self-supervised learning without labeled data
• Analytical regularization parameter selection