graph neural networks

Learning to Explain Solutions of Optimal Control Problems

The results show that the GNN model can accurately predict the optimal values of the manipulated variables, and application of explainable AI algorithms reveals equality and inequality constraints that are the most important for predicting the optimal solution.

Meal-Induced Proton Density Fat Fraction and T 2 ∗ Decrease in Supraclavicular Adipose Tissue.

Brown adipose tissue (BAT) is a metabolically active tissue in humans, located primarily within the supraclavicular adipose tissue (scAT), that can be activated by cold or high-caloric meal consumption. While the changes of proton density fat fraction (PDFF) upon cold activation are well investigated, there is a knowle...

FOSLS-deRhaNN: native de Rham neural classes for H(div) and H(curl) with applications to first-order system least-squares neural network methods for partial differential equations

We construct neural approximation classes native to the graph spaces H(div) and H(curl), in two and three dimensions and, for H(div), in any dimension. Every realization lies in the space for all parameter values, and with kinked potentials, such as ReLU networks, the admissible jumps appear at finite width. The classe...

Graph Neural Network-Driven Deep Reinforcement Learning for Scalable RIS Allocation

A scalable framework combining Graph Neural Networks (GNNs) with Deep Reinforcement Learning (DRL) for dynamic shared RIS orchestration and introduces a physical topology sparsification strategy that prunes dense channel matrices into a sparse tripartite graph, improving global coverage probability while reducing compu...

Interpretable Multimodal Engagement Prediction with Graph-based Mixture-of-Experts

A graph-based Mixture-of-Experts (MoE) framework that explicitly separates self and social influences to enable interpretable engagement modeling, which outperforms baselines by up to 64.1% while offering interpretable insights into engagement dynamics across language and gender groups.

Attention-enhanced graph neural networks for nonlinearity compensation in optical fiber communication

This paper presents AEGNN (attention-enhanced graph neural network), framework that integrated Graph Attention Networks (GAT) with multi-head self-attention mechanisms to model space-time topology of optical networks and is the first fully reproducible framework combining attention-based GNNs with open-source optical c...

Causal Graph Constrained LLM for Fault Diagnosis in Industrial Data Centers

Industrial data-center systems contain complex component dependencies and long fault-propagation chains, making accurate root-cause localization difficult. Large language models (LLMs) provide a promising solution because of their strong ability to understand, organize, and reason over heterogeneous operational evidenc...

Graph Neural Network-Based Analysis for Digital Logic Circuit Optimization

Abstract Digital logic circuits are becoming increasingly complex, making efficient circuit optimization an important part of Electronic Design Automation (EDA). Traditional circuit optimization methods may require extensive computation and predefined rules when dealing with complex circuit structures. This study aims...

Intelligent Digital Logic Circuit Optimization Through Graph Neural Networks

Abstract The increasing complexity of digital logic circuits creates a need for effective methods of circuit analysis and optimization. Conventional Electronic Design Automation (EDA) techniques commonly depend on established algorithms and predefined rules, which can become difficult to apply to more complicated circu...

UEXGNN: reference inference release of a physically constrained autoregressive graph neural network for transient-field prediction of near-seabed underwater explosions

Strain-Engineered Magnetism, Super-Exchange Physics, and Thermal Phase Transitions in Monolayer Chromium Trihalides CrX3 (X = Cl, Br, I)

This repository contains the Python calculation scripts, raw data JSON files, and publication-quality vector figures supporting the manuscript: "Strain-Engineered Magnetism, Super-Exchange Physics, and Thermal Phase Transitions in Monolayer Chromium Trihalides CrX3 (X = Cl, Br, I)" ### Repository Structure & Contents:...

Smart Circuit Structure Learning for Digital Logic Optimization

Abstract The increasing scale and interconnectedness of digital logic designs make it more difficult to examine their internal organization and determine where improvements can be made. Conventional Electronic Design Automation (EDA) methods generally depend on established algorithms and rule-driven procedures, which m...