High-Performance and Programmable Attentional Graph Neural Networks with Global Tensor Formulations.
Maciej BestaPawel RencRobert GerstenbergerPaolo Sylos LabiniAlexandros Nikolaos ZiogasTiancheng ChenLukas GianinazziFlorian ScheidlKalman SzenesArmon CarigietPatrick IffGrzegorz KwasniewskiRaghavendra KanakagiriChio GeSammy JaegerJaroslaw WasFlavio VellaTorsten HoeflerPublished in: SC (2023)
Keyphrases
- neural network
- pattern recognition
- fuzzy logic
- back propagation
- graph model
- artificial neural networks
- random walk
- graph representation
- high order
- genetic algorithm
- signal processor
- tensor space
- global consistency
- weighted graph
- connected components
- higher order
- directed graph
- directed acyclic graph
- visual search
- undirected graph
- fault diagnosis
- learning algorithm
- neural network model
- selective attention
- global structure
- low cost
- graph structure
- graph theory
- bipartite graph
- neural nets
- optimization methods