Researchers have developed Fed-PIGANO, a federated learning framework designed to train physics-informed geometry-aware neural operator (PI-GANO) models for additive manufacturing. This approach allows individual manufacturing sites to train models on their proprietary part geometries without sharing the data. The framework aggregates geometry encoders and prediction networks, enabling accurate thermal field predictions on unseen parts. Fed-PIGANO demonstrated strong performance on a 2D thermal benchmark, achieving low relative L2 errors and comparable accuracy to centralized models, thus facilitating rapid thermal prediction for design optimization and inverse design tasks. AI
IMPACT Enables distributed manufacturing sites to collaborate on AI model training without compromising proprietary design data.
RANK_REASON The cluster contains an academic paper detailing a new machine learning framework. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →