FPGA implementation of convolutional neural network based on stochastic computing

Kim, Dae Woo; Moghaddam, Mansureh S.; Moradian, Hossein; Sim, Hyeonuk; Lee, Jong-Eun; Choi, Ki‐Young

doi:10.1109/fpt.2017.8280162

Cited by 7 publications

(1 citation statement)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A further interesting approach is found in [37], where SC components on an FPGA process sensor data in a fault-tolerant manner. Kim et al [38] presented an implementation which maps parts of a convolutional NN to the FPGA's LUTs while using the on-FPGA microprocessor for the remaining operations. In [39], SC-enabled modules accelerate stochastic gradient descent computations during deep learning.…”

Section: Case Studies and Practical Implementationsmentioning

confidence: 99%

Guest Editors’ Introduction: Stochastic Computing for Neuromorphic Applications

et al. 2021

View full text Add to dashboard Cite

This extended editorial provides an introduction into stochastic computing (SC) and its usage in emerging neuromorphic applications. It covers SC primitives and the tradeoffs that occur when designing larger SC-based systems. The article shows how SC enables low-cost, low-power, and errortolerant hardware implementation of neural networks suitable for edge computing. It provides a brief survey about recent proposals in this domain and introduces the articles in this special issue.-Jörg Henkel, Karlsruhe Institute of Technology  After decAdes of research on general-purpose computing, the main pathway of computer architecture research has recently shifted to domain-specific concepts. In their Turing lecture in 2018, Hennessy and Patterson [1] have called the transition to domain-specific languages and architectures a "golden age for computer architects." Neuromorphic architectures have raised tremendous interest from researchers and industrial users. We can distinguish two main trends: specialized neural network (NN) processors with size and throughput

show abstract

Section: Case Studies and Practical Implementationsmentioning

confidence: 99%