Feature-Aware Anisotropic Local Differential Privacy for Utility-Preserving Graph Representation Learning in Metal Additive Manufacturing

2026-04-06

AI Summary

This work addresses the challenges in defect detection for metal additive manufacturing, where high-quality sensor data are rarely shared due to proprietary concerns and existing models neglect inter-layer physical couplings. To overcome the utility degradation caused by uniform noise injection in conventional local differential privacy (LDP) mechanisms, the authors propose FI-LDP-HGAT—a non-interactive privacy-preserving framework that integrates a hierarchical graph attention network to model spatial thermal couplings between scan paths and deposited layers. A feature-aware anisotropic Gaussian mechanism is introduced, allocating privacy budgets according to feature importance to inject less noise into critical thermal signals while maintaining strict LDP guarantees. Experiments on a directed energy deposition (DED) porosity dataset show that the method achieves 81.5% utility recovery at ε=4 and a defect recall of 0.762 at ε=2, significantly outperforming baseline approaches. Moreover, feature importance exhibits a strong negative correlation with injected noise levels (Spearman’s ρ = −0.81).

Abstract

Metal additive manufacturing (AM) enables the fabrication of safety-critical components, but reliable quality assurance depends on high-fidelity sensor streams containing proprietary process information, limiting collaborative data sharing. Existing defect-detection models typically treat melt-pool observations as independent samples, ignoring layer-wise physical couplings. Moreover, conventional privacy-preserving techniques, particularly Local Differential Privacy (LDP), lead to severe utility degradation because they inject uniform noise across all feature dimensions. To address these interrelated challenges, we propose FI-LDP-HGAT. This computational framework combines two methodological components: a stratified Hierarchical Graph Attention Network (HGAT) that captures spatial and thermal dependencies across scan tracks and deposited layers, and a feature-importance-aware anisotropic Gaussian mechanism (FI-LDP) for non-interactive feature privatization. Unlike isotropic LDP, FI-LDP redistributes the privacy budget across embedding coordinates using an encoder-derived importance prior, assigning lower noise to task-critical thermal signatures and higher noise to redundant dimensions while maintaining formal LDP guarantees. Experiments on a Directed Energy Deposition (DED) porosity dataset demonstrate that FI-LDP-HGAT achieves 81.5% utility recovery at a moderate privacy budget (epsilon = 4) and maintains defect recall of 0.762 under strict privacy (epsilon = 2), while outperforming classical ML, standard GNNs, and alternative privacy mechanisms, including DP-SGD across all evaluated metrics. Mechanistic analysis confirms a strong negative correlation (Spearman = -0.81) between feature importance and noise magnitude, providing interpretable evidence that the privacy-utility gains are driven by principled anisotropic allocation.

Problem

• Local Differential Privacy
• Graph Representation Learning
• Metal Additive Manufacturing
• Utility Preservation
• Feature-aware Privacy

Innovation

• Feature-Aware Anisotropic LDP
• Hierarchical Graph Attention Network
• Utility-Preserving Privacy
• Metal Additive Manufacturing
• Non-Interactive Feature Privatization