Polymorphic Accelerators for Deep Neural Networks

Azizimazreah, Arash; Chen, Lizhong

doi:10.1109/tc.2020.3048624

Cited by 10 publications

(3 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Network. e artificial neural network (abbreviated as ANN) [23] is a complex network composed of a large number of simple processing units widely connected, which is used to simulate the working mode of the human brain. It reflects the basic functions of many human brains, but it is not a true portrayal of all human brains, but only with some simplification, abstraction, and simulation.…”

Section: New Perspective Of Credit Risk Assessment: Neural Network Me...mentioning

confidence: 99%

Application of Business Intelligence Based on the Deep Neural Network in Credit Scoring

Feng

Chen

2022

Security and Communication Networks

View full text Add to dashboard Cite

In order to solve the economic globalization and the intensification of financial market volatility, this study puts forward the judgment of credit risk of loan enterprises, so as to effectively improve the ability of commercial banks to control credit risk. At the same time, neural network technology is applied to solve nonlinear data modeling. Experiments and evaluation are carried out by using the artificial neural network (ANN) evaluation model. The experimental evaluation model data obtained from the experimental evaluation model show that the prediction effect is good, which proves the effectiveness of this method. According to the data of China Banking Regulatory Commission, at the end of the third quarter of 2016, the balance of nonperforming loans of Chinese commercial banks was 1493.9 billion yuan. By the end of the third quarter of 2021, the balance of nonperforming loans of Chinese commercial banks had reached 28153.5 trillion yuan, a month on month increase of about 42.7 billion yuan. At the end of the third quarter of 2021, the domestic and foreign currency assets of Chinese banking financial institutions were 339.4 trillion yuan, a year-on-year increase of 7.7%. This shows that the credit risk remains high, but it is still at a controllable level. Commercial banks are the center of the modern economic system. Their credit risk is very important to the stability and safe development of national economy. It is also the key factor for its own business sustainability and the protection of the interests of depositors. Therefore, effectively preventing and reducing credit risk is a key field in the financial risk management of commercial banks.

show abstract

Section: New Perspective Of Credit Risk Assessment: Neural Network Me...mentioning

confidence: 99%

Application of Business Intelligence Based on the Deep Neural Network in Credit Scoring

Feng

Chen

2022

Security and Communication Networks

View full text Add to dashboard Cite

show abstract

“…For example, the latency of MRAM tends to be substantially larger than that of SRAM latency [58]. Moreover, the bandwidth of local memory also varies between memory blocks, depending on the number of banks allocated to those blocks (e.g., [19] and [59,60]). Thus, each PE tends to experience an order-of-magnitude difference in its latency and bandwidth, depending on which memory block the activations (or filters) are transferred from/to.…”

Section: Spatial Data Dependence Graphmentioning

confidence: 99%

Spatial Data Dependence Graph Based Pre-RTL Simulator for Convolutional Neural Network Dataflows

Wang

Park

2022

IEEE Access

View full text Add to dashboard Cite

In this paper, a new pre-RTL simulator is proposed to predict the power, performance, and area of convolutional neural network (CNN) dataflows prior to register-transfer-level (RTL) design. In the simulator, a novel approach is adopted to implement a spatial data dependence graph (SDDG), which enables us to model a specific dataflow alongside inter-instruction dependencies by tracking the status of each processing element (PE). In addition, the proposed pre-RTL simulator makes it possible to evaluate the impact of memory constraints such as latency and bandwidth. The latency-insensitive and bandwidth-insensitive PE controllers assumed in the proposed pre-RTL simulator guarantee both operational correctness and maximum performance, regardless of memory constraints. In particular, it is shown that the optimal distribution method of local memory bandwidth can reduce the accelerator performance by up to 37.6% compared with the equal distribution method. For weight stationary (WS) and row stationary (RS) dataflows, the accelerator performance closely depends on memory constraints. The simulation results also show that the relative performances of dataflows depend on the layer shape of the convolutional layer. For example, for an identical hardware area in a standard convolutional layer of AlexNet, WS dataflows do not provide any performance gain over RS dataflows when the memory latency is sufficiently high. In addition, WS dataflows cannot fully reuse the input activation, thereby increasing local memory accesses, since the number of weights loaded at a specific time is limited. Moreover, in a depth-wise convolutional layer of MobileNet, WS dataflows tend to outperform RS dataflows even in the presence of large memory latency.

show abstract

“…[24][25][26][27] However, the state-of-the-art (SOTA) in the literature reveals various challenges in designing MAC operators within constrained environments. These challenges include: (I) inefficient utilization of HW resources due to static reconfiguration approaches; 23,[28][29][30] (II) limitations imposed by static precision regarding overflow and underflow management; 26,27,31 (III) computation integrity due to static precision regarding overflow or underflow occurrence.…”

Section: Introductionmentioning

confidence: 99%

Self-adapting reconfigurable multiply-accumulator for real-time image processing in embedded systems

Fasolino,

Vitolo,

Liguori

et al. 2024

Real-Time Processing of Image, Depth, and Video Information 2024

View full text Add to dashboard Cite

Multiply-Accumulate (MAC) operation is widely used in various real-time image processing tasks, ranging from Convolutional Neural Networks to digital filtering, significantly impacting overall system performance. In this work the Self-Adapting Reconfigurable Multiply-Accumulate (SR-MAC) is proposed as a new instrument to find the optimal trade-off between operation throughput, power consumption and physical resources utilization in real-time image processing applications. Operations of the proposed system rely on the dynamic reconfiguration of the hardware resources on the basis of the current computational requirements. This is achieved by monitoring overflow and over-representation occurrences at each accumulation cycle, and properly considering the relevant portion of the accumulation result. A custom architecture of the proposed algorithm has been designed and implemented on an AMD Xilinx Artix-7 FPGA through a Verilog description and compared to the AMD Xilinx fixed-point macro (floating-point fused multiply-accumulate). The SR-MAC achieves reductions of 83% (82%), 79% (93%) and 87.2% (94%) in the number of LUTs, FFs, and the power dissipation, P dynN , respectively. The SR-MAC has also been used to replace arithmetic units in typical real-time image processing applications. In these cases, its employment has allowed the reduction up to 6% and 14% of FFs and P dynN , respectively, while increasing up to 14% the f M ax . These results highlight the significant performance enhancement achieved with respect to both single operators and entire systems, making SR-MAC an excellent design choice in real-time image processing applications.

show abstract

Polymorphic Accelerators for Deep Neural Networks

Cited by 10 publications

References 35 publications

Application of Business Intelligence Based on the Deep Neural Network in Credit Scoring

Application of Business Intelligence Based on the Deep Neural Network in Credit Scoring

Spatial Data Dependence Graph Based Pre-RTL Simulator for Convolutional Neural Network Dataflows

Self-adapting reconfigurable multiply-accumulator for real-time image processing in embedded systems

Contact Info

Product

Resources

About