Paper Info
Title
Initial Study Of Reconfigurable Neural Network Accelerators
Abstract
Neural Networks or NNs are widely used for many machine learning applications such as image processing and speech recognition. Since general-purpose processors such as CPUs and GPUs are energy inefficient for computing NNs, application-specific hardware accelerators for NNs (a.k.a. Neural Network Accelerators or NNAs) have been proposed to improve the energy efficiency. However, the existing NNAs are too customized for computing specific NNs, and do not allow to change neuron models or learning algorithms. This limitation prevents machine-learning researchers from exploiting NNAs, so we are developing a general-purpose NNA including reconfigurable logic, which is called a reconfigurable NNA or RNNA. The RNNA is highly tuned for the NN computation but allows end users to customize the hardware to compute desired NNs. This paper introduces the RNNA architecture, and reports the performance analysis of the RNNA with an in-house cycle-level simulator.
Year
DOI
Venue
2016
10.1109/CANDAR.2016.106
2016 FOURTH INTERNATIONAL SYMPOSIUM ON COMPUTING AND NETWORKING (CANDAR)
Field
DocType
ISSN
Computer architecture,End user,Efficient energy use,Computer science,Parallel computing,Image processing,Artificial neural network,Computation
Conference
2379-1888
Citations 
PageRank 
References 
0
0.34
0
Authors
6
Name
Order
Citations
PageRank
Momoka Ohba100.34
Satoshi Shindo200.34
Shinobu Miwa32813.09
Tomoaki Tsumura45814.63
Hayato Yamaki503.04
Hiroki Honda621.09