Predicting DNN run-times using a Graph Neural Network for heterogenous SoCs.
Andrei VimanMihai NegruCatalin CasonPublished in: ICCP (2023)
Keyphrases
- run times
- neural network
- graph representation
- training process
- random walk
- artificial neural networks
- graph theoretic
- weighted graph
- graph structure
- graph theory
- connected components
- pattern recognition
- structured data
- neural network model
- neural network is trained
- graph based algorithm
- graph mining
- multi layer
- feed forward
- directed graph
- back propagation
- data sets
- network architecture
- spanning tree
- graph partitioning
- fuzzy logic
- bipartite graph
- varying degrees
- dependency graph
- fuzzy artmap
- genetic algorithm