DyGNN: Algorithm and Architecture Support of Dynamic Pruning for Graph Neural Networks.
Cen ChenKenli LiXiaofeng ZouYangfan LiPublished in: DAC (2021)
Keyphrases
- neural network
- graph structure
- learning algorithm
- graph based algorithm
- detection algorithm
- objective function
- computational complexity
- preprocessing
- search space
- cost function
- dynamic programming
- pattern recognition
- simulated annealing
- particle swarm optimization
- optimization algorithm
- maximum flow
- graph model
- optimal solution
- spanning tree
- matching algorithm
- tree structure
- random walk
- bipartite graph
- dynamic environments
- weighted graph
- hardware implementation
- graphical models
- minimum spanning tree
- pruning strategy
- graph clustering
- np hard
- decision trees