iMAT: Energy-Efficient In-Memory Acceleration for Ternary Neural Networks With Sparse Dot Product.
Shien ZhuShuo HuaiGuochu XiongWeichen LiuPublished in: ISLPED (2023)
Keyphrases
- energy efficient
- dot product
- neural network
- wireless sensor networks
- positive semi definite
- sensor networks
- energy consumption
- similarity function
- kernel function
- gaussian kernels
- high dimensional
- base station
- energy efficiency
- feature space
- associative memory
- multi core architecture
- routing protocol
- back propagation
- routing algorithm
- data transmission
- similarity measure
- kernel methods
- power consumption
- sensor data
- pairwise