Title | ||
---|---|---|
A Configurable Architecture of ANN in Hardware with Resource-Efficient Reusable Neuron. |
Abstract | ||
---|---|---|
Artificial Neural Network (ANN) is one of the most important structure of machine learning and it has been widely used in various areas such as medical diagnosis, image classification and signal processing. The large area and resource cost make it difficult for the realization of ANN in hardware. This paper presents a flexible architecture of ANN with resource-efficient reusable neuron. The dynamical activation method of neurons is proposed and utilized to make the ANN more flexible and configurable. The designed structure of neuron supports multiple calculation modes which is reused in both feedforward and back-propagation. The approach of multiplexing layer is adopted to reduce the number of physical layers to 1. The simulation results show the proposed method can achieve the function of ANN while significantly reducing the resource cost and area. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/ASICON47005.2019.8983505 | ASICON |
Field | DocType | Citations |
Signal processing,Architecture,Computer science,Computer hardware,Artificial neural network,Multiplexing,Contextual image classification,Medical diagnosis,Feed forward | Conference | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jiahao Lu | 1 | 4 | 3.09 |
Xianghua Luo | 2 | 0 | 0.34 |
dongsheng liu | 3 | 15 | 6.85 |
Peng Liu | 4 | 23 | 8.57 |
Bo Liu | 5 | 0 | 0.34 |