Hybrid Convolution Architecture for Energy-Efficient Deep Neural Network Processing

Kim, Suchang; Jo, Jihyuck; Park, In-Cheol

doi:10.1109/tcsi.2021.3059882

Cited by 5 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DNN accelerators have been developed with various design approaches [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. Due to the data-centric property in recent ASIC-based DNN accelerators, in which a significantly large amount of data should be processed and transferred in and out of the accelerator chips, memory plays an important role.…”

Section: Dnn Acceleratorsmentioning

confidence: 99%

Dataflow Optimization through Exploring Single-Layer and Inter-Layer Data Reuse in Memory-Constrained Accelerators

Ye¹,

Yanagisawa

Shi

2022

Electronics

View full text Add to dashboard Cite

Off-chip memory access has become the performance and energy bottleneck in memory-constrained neural network accelerators. To provide a solution for the energy efficient processing of various neural network models, this paper proposes a dataflow optimization method for modern neural networks by exploring the opportunity of single-layer and inter-layer data reuse to minimize the amount of off-chip memory access in memory-constrained accelerators. A mathematical analysis of three inter-layer data reuse methods is first presented. Then, a comprehensive exploration to determine the optimal data reuse strategy from single-layer and inter-layer data reuse approaches is proposed. The result shows that when compared to the existing single-layer-based exploration method, SmartShuttle, the proposed approach can achieve up to 20.5% and 32.5% of off-chip memory access reduction for ResNeXt-50 and DenseNet-121, respectively.

show abstract

Section: Dnn Acceleratorsmentioning

confidence: 99%

Dataflow Optimization through Exploring Single-Layer and Inter-Layer Data Reuse in Memory-Constrained Accelerators

Ye¹,

Yanagisawa

Shi

2022

Electronics

View full text Add to dashboard Cite

show abstract

“…Accelerating deep neural network processing in edge computing using energy-efficient platforms is an important goal [ 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ]. Currently, most object detection and classification models are carried out in graphics processing units.…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, many lightweight approaches with low-power consumption and low-computational performance have emerged recently. A few dedicated neural network accelerators have been implemented on FPGA hardware platforms [ 12 , 14 , 17 , 21 , 23 ], while several authors proposed ASIC-based neural network accelerators [ 13 , 15 , 16 , 18 , 19 , 22 ]. Samimi et al [ 20 ] proposed a technique based on the residue number system to improve the energy efficiency of deep neural network processing.…”

Section: Introductionmentioning

confidence: 99%

An Approximate GEMM Unit for Energy-Efficient Object Detection

Pilipović

Risojević

Božič

et al. 2021

Sensors

View full text Add to dashboard Cite

Edge computing brings artificial intelligence algorithms and graphics processing units closer to data sources, making autonomy and energy-efficient processing vital for their design. Approximate computing has emerged as a popular strategy for energy-efficient circuit design, where the challenge is to achieve the best tradeoff between design efficiency and accuracy. The essential operation in artificial intelligence algorithms is the general matrix multiplication (GEMM) operation comprised of matrix multiplication and accumulation. This paper presents an approximate general matrix multiplication (AGEMM) unit that employs approximate multipliers to perform matrix–matrix operations on four-by-four matrices given in sixteen-bit signed fixed-point format. The synthesis of the proposed AGEMM unit to the 45 nm Nangate Open Cell Library revealed that it consumed only up to 36% of the area and 25% of the energy required by the exact general matrix multiplication unit. The AGEMM unit is ideally suited to convolutional neural networks, which can adapt to the error induced in the computation. We evaluated the AGEMM units’ usability for honeybee detection with the YOLOv4-tiny convolutional neural network. The results implied that we can deploy the AGEMM units in convolutional neural networks without noticeable performance degradation. Moreover, the AGEMM unit’s employment can lead to more area- and energy-efficient convolutional neural network processing, which in turn could prolong sensors’ and edge nodes’ autonomy.

show abstract