As a key ingredient of deep neural networks (DNNs), fully-connected (FC) layers are widely used in various artificial intelligence applications. However, there are many parameters in FC layers, so the efficient process of FC layers is restricted by memory bandwidth. In this paper, we propose a compression approach combining block-circulant matrix-based weight representation and power-of-two quantization. Applying block-circulant matrices in FC layers can reduce the storage complexity from O ( k 2 ) to O ( k ) . By quantizing the weights into integer powers of two, the multiplications in the reference can be replaced by shift and add operations. The memory usages of models for MNIST, CIFAR-10 and ImageNet can be compressed by 171 × , 2731 × and 128 × with minimal accuracy loss, respectively. A configurable parallel hardware architecture is then proposed for processing the compressed FC layers efficiently. Without multipliers, a block matrix-vector multiplication module (B-MV) is used as the computing kernel. The architecture is flexible to support FC layers of various compression ratios with small footprint. Simultaneously, the memory access can be significantly reduced by using the configurable architecture. Measurement results show that the accelerator has a processing power of 409.6 GOPS, and achieves 5.3 TOPS/W energy efficiency at 800 MHz.
In this work, artificial neural network (ANN) models and particle swarm optimization (PSO) models based on machine learning were built to predict the HHVs of MSW quickly. Four kinds of BP ANN models and two PSO models were built using proximate analysis and ultimate analysis of 33 MSW samples as input variables. As a comparison, three classical linear empirical models employed from publications were also used. The modeling results show that the input variables had significant influence on the prediction accuracy. The ANN model based on proximate analysis had lower precision for HHV estimation. The ultimate analysis had better prediction performance while the combination of ultimate analysis and proximate analysis (ANN-4 model) had the best accuracy. With regard to ANN-4 model, the largest relative deviation for all samples was lower than 10%. Due to the complex composition of MSW, the linear empirical model was not suitable for accurate prediction of the calorific value of MSW. Nonlinear empirical formulas obtained by PSO models improved the prediction performance for most samples. In general, the ANN modeling method could predict the thermochemical properties of MSW and provide rapid and effective guidance for the operation of MSW incineration process.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.