GNN Computation

Graph neural network (GNN) computation focuses on efficiently processing graph-structured data using deep learning techniques, aiming to improve both the speed and accuracy of GNN models. Current research emphasizes optimizing GNN algorithms, such as graph convolutional networks and graph attention networks, through techniques like novel aggregation methods, hardware acceleration (including specialized accelerators and optimized use of existing hardware like GPUs), and quantization to reduce computational cost. These advancements are crucial for enabling the application of GNNs to large-scale problems in diverse fields, including personalized medicine, biological systems analysis, and social network analysis, where their ability to model complex relationships is highly valuable.

Papers

May 31, 2024

G-Transformer for Conditional Average Potential Outcome Estimation over Time
Konstantin Hess, Dennis Frauen, Valentyn Melnychuk, Stefan Feuerriegel
End to End Time Matter Conditional Average Treatment Effect GNN Computation Personalized Decision

May 26, 2024

Integrating GNN and Neural ODEs for Estimating Two-Body Interactions in Mixed-Species Collective Motion
Masahito Uwamichi, Simon K. Schnyder, Tetsuya J. Kobayashi, Satoshi Sawai
Neural ODE Interaction Strategy Collective Behavior Complex Dynamic Collective Motion GNN Computation Cell Migration

April 23, 2024

NeuraChip: Accelerating GNN Computations with a Hash-based Decoupled Spatial Accelerator
Kaustubh Shivdikar, Nicolas Bohm Agostini, Malith Jayaweera, Gilbert Jonatan, Jose L. Abellan, Ajay Joshi, John Kim, David Kaeli
Graph Neural Network Large Scale Graph Datasets GNN Computation GNN Acceleration Spatial Accelerator

December 16, 2023

Enabling Accelerators for Graph Computing
Kaustubh Shivdikar
Graph Neural Network Gene Level GNN Multi Accelerator DNN Workload GNN Computation

October 18, 2023

Architectural Implications of GNN Aggregation Programming Abstractions
Yingjie Qi, Jianlei Yang, Ao Zhou, Tong Qiao, Chunming Hu
Graph Neural Network GNN Framework GNN Aggregation GNN Acceleration GNN Computation

August 23, 2023

Cached Operator Reordering: A Unified View for Fast GNN Training
Julia Bazinska, Andrei Ivanov, Tal Ben-Nun, Nikoli Dryden, Maciej Besta, Siyuan Shen, Torsten Hoefler
Graph Neural Network Graph Convolutional Network Unified View GNN Training GNN Layer GNN Computation Large Scale GNN

August 4, 2023

Exploiting On-chip Heterogeneity of Versal Architecture for GNN Inference Acceleration
Paul Chen, Pavan Manjunath, Sasindu Wijeratne, Bingyi Zhang, Viktor Prasanna
Graph Convolutional Network GNN Inference GNN Computation

August 2, 2023

Tango: rethinking quantization for graph neural network training on GPUs
Shiyang Chen, Da Zheng, Caiwen Ding, Chengying Huan, Yuede Ji, Hang Liu
Neural Network Graph Neural Network Graph Drawing Single GPU Multiplier Free Quantization GNN Training GNN Computation

June 27, 2023

SENSEi: Input-Sensitive Compilation for Accelerating GNNs
Damitha Lenadora, Vimarsh Sathia, Gerasimos Gerogiannis, Serif Yesil, Josep Torrellas, Charith Mendis
Graph Neural Network Graph Attention Network GNN Model GNN Based Service Sparse Matrix Multiplication Efficient Compilation GNN Computation

October 20, 2022

gSuite: A Flexible and Framework Independent Benchmark Suite for Graph Neural Network Inference on GPUs
Taha Tekdoğan, Serkan Göktaş, Ayse Yilmazer-Metin
Graph Neural Network Single GPU Extension Study GNN Model GNN Framework GNN Computation Graph Neural Network Inference

May 19, 2022

Parallel and Distributed Graph Neural Networks: An In-Depth Concurrency Analysis
Maciej Besta, Torsten Hoefler
Graph Neural Network GNN Model GNN Performance GNN Computation