An Efficient Sparse LSTM Accelerator on Embedded FPGAs with Bandwidth-Oriented Pruning.
Shiqing LiShien ZhuXiangzhong LuoTao LuoWeichen LiuPublished in: FPL (2023)
Keyphrases
- embedded systems
- field programmable gate array
- search space
- reverse k nearest neighbor
- smart camera
- sparse representation
- low cost
- computationally efficient
- sparse data
- early vision
- hardware implementation
- effective pruning
- avoid overfitting
- bandwidth allocation
- hardware software
- parallel implementation
- data mining
- recurrent neural networks
- dimensionality reduction
- high dimensional
- feature space