GROW: A Row-Stationary Sparse-Dense GEMM Accelerator for Memory-Efficient Graph Convolutional Neural Networks.
Minhoo KangRanggi HwangJiwon LeeDongyun KamYoungjoo LeeMinsoo RhuPublished in: CoRR (2022)
Keyphrases
- memory efficient
- convolutional neural networks
- graph search
- densely connected
- non stationary
- external memory
- iterative deepening
- dense stereo
- convolutional network
- multiple sequence alignment
- gaussian graphical models
- quasi cliques
- weighted graph
- graph representation
- integral image
- parallel implementation
- binary matrices
- sparse representation
- dense optical flow
- random walk
- directed acyclic graph
- graph theory
- graph matching
- high dimensional
- dense sampling
- graph mining
- dense motion estimation