Proceedings of the AAAI Conference on Artificial Intelligence · 2018 · 31 citations · 11 references
ComputingComputer EngineeringComputer ScienceArtificial IntelligenceComputer ArchitectureMachine LearningAccelerator ArchitectureAccelerator IntegrationDeep LearningHardware AccelerationNeural NetworksPattern RecognitionAcceleratorsClassification AccuracyClassification Accuracy AnalysisConvolution Engine DesignDeep Learning AccelerationDeep Learning AcceleratorsEnduranceFloating-point Unit Design
The heavy burdens of computation and off-chip traffic impede deploying the large scale convolution neural network on embedded platforms. As CNN is attributed to the strong endurance to computation errors, employing block floating point (BFP) arithmetics in CNN accelerators could save the hardware cost and data traffics efficiently, while maintaining the classification accuracy. In this paper, we verify the effects of word width definitions in BFP to the CNN performance without retraining. Several typical CNN models, including VGG16, ResNet-18, ResNet-50 and GoogLeNet, were tested in this paper. Experiments revealed that 8-bit mantissa, including sign bit, in BFP representation merely induced less than 0.3% accuracy loss. In addition, we investigate the computational errors in theory and develop the noise-to-signal ratio (NSR) upper bound, which provides the promising guidance for BFP based CNN engine design.
11
ImageNet Large Scale Visual Recognition Challenge
Olga Russakovsky, Jia Deng, Hao Su et al. · International Journal of Computer Vision · 2015 · 40.2K citations
Optimizing FPGA-based Accelerator Design for Deep Convolutional Neural Networks
Chen Zhang, Peng Li, Guangyu Sun et al. · 2015 · 2.1K citations
Accelerating Deep Convolutional Neural Networks Using Specialized Hardware
Kalin Ovtcharov, Olatunji Ruwase, Joo-Young Kim et al. · 2015 · 315 citations