SIAM: Chiplet-based Scalable In-Memory Acceleration with Mesh for Deep Neural Networks.
Gokul KrishnanSumit K. MandalManvitha PannalaChaitali ChakrabartiJae-sun SeoÜmit Y. OgrasYu CaoPublished in: CoRR (2021)
Keyphrases
- neural network
- auto associative
- low memory
- associative memory
- pattern recognition
- artificial neural networks
- limited memory
- d mesh
- fuzzy logic
- memory requirements
- neural network model
- highly scalable
- computing power
- network architecture
- memory size
- neural nets
- recurrent neural networks
- computational power
- activation function
- multilayer perceptron
- feed forward
- statistical analysis and data mining
- web scale
- back propagation
- information processing
- genetic algorithm
- memory usage
- memory space
- memory efficient
- main memory