FPGA implementation of neuron block for artificial neural network

Li, Zhaofang; Huang, Yu-Jung; Lin, Wei-Cheng

doi:10.1109/edssc.2017.8126431

Cited by 9 publications

(6 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The utilization of neural networks in FPGAs involves employing smaller networks with a complete connection configuration [ 27 , 28 ]. When implementing LSTM networks, individual neurons are constructed as separate blocks [ 33 ].…”

Section: Reduction Of Lstm Network Parametersmentioning

confidence: 99%

“…The main advantage of using FPGA cards is that they can be used to implement massively parallel data-processing algorithms, through a programmable structure based on blocks that operate together, allowing a reconfiguration if necessary to improve connections from stage to stage, necessary for a specific algorithm [27,28]. Microcontrollers are another tool used, but they work sequentially, limiting their development process and application to already-defined tasks [27] FPGA boards contain logic gates, clock controllers, and RAM and ROM memories, among other elements.…”

Section: Fpga Boardmentioning

confidence: 99%

Section: Fpga Boardmentioning

confidence: 99%

“…Field programmable gate arrays (FPGAs) are programmable electronic devices based on a matrix of logic blocks whose interconnection and functionality can be configured by the user using a specialized description language (such as VHDL and Verilog) [ 25 , 26 ], belonging to a classification of an embedded system [ 27 ]. The term embedded system refers to small systems with limited resources, referencing the operability of a personal computer [ 25 , 26 , 27 ].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Modeling of Particulate Pollutants Using a Memory-Based Recurrent Neural Network Implemented on an FPGA

Ramírez-Montañez,

Rangel-Magdaleno,

Aceves-Fernández

et al. 2023

Micromachines

View full text Add to dashboard Cite

The present work describes the training and subsequent implementation on an FPGA board of an LSTM neural network for the modeling and prediction of the exceedances of criteria pollutants such as nitrogen dioxide (NO2), carbon monoxide (CO), and particulate matter (PM10 and PM2.5). Understanding the behavior of pollutants and assessing air quality in specific geographical regions is crucial. Overexposure to these pollutants can cause harm to both natural ecosystems and living organisms, including humans. Therefore, it is essential to develop a solution that can accurately evaluate pollution levels. One potential approach is to implement a modified LSTM neural network on an FPGA board. This implementation obtained an 11% improvement compared to the original LSTM network, demonstrating that the proposed architecture is able to maintain its functionality despite reducing the number of neurons in its initial layers. It shows the feasibility of integrating a prediction network into a limited system such as an FPGA board, but easily coupled to a different system. Importantly, this implementation does not compromise the prediction accuracy for both 24 h and 72 h time frames, highlighting an opportunity for further enhancement and refinement.

show abstract

Section: Reduction Of Lstm Network Parametersmentioning

confidence: 99%

Section: Fpga Boardmentioning

confidence: 99%

Section: Fpga Boardmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Modeling of Particulate Pollutants Using a Memory-Based Recurrent Neural Network Implemented on an FPGA

Ramírez-Montañez,

Rangel-Magdaleno,

Aceves-Fernández

et al. 2023

Micromachines

View full text Add to dashboard Cite

show abstract

“…The two-step approach is reported to have an improved accuracy that is 10 times better than that of using only SONF and twice better than just using LUT. Li et al [25] implemented a neuron block with sigmoid function using the CORDIC algorithm. Some of the applications of ANN implementation on FPGA include classification of the region of pixels i.e., hand regions by Krips et al [26] using three inputs representing RGB values with a single hidden layer and one output with data represented by integers and weights scaled up and rounded off to the nearest integer.…”

Section: Introductionmentioning

confidence: 99%

Neural net implementation of steam properties on FPGA

Dutt¹,

Ganesh²,

Premchand

2021

IJRES

View full text Add to dashboard Cite

Real time applications like model predictive control, monitoring and data reconciliation of power plants and industrial processes employ nonlinear mathematical models and require thermodynamic properties and their derivatives of working fluids. Applications like super heater temperature control based on energy balance and real time data reconciliation, require an efficient and a compact method for simultaneous estimation of thermodynamic properties, and their partial derivatives suitable for implementation in field-programmable gate array (FPGA). However, the complex mathematical formulations of these properties prohibit direct implementations in FPGAs. Single artificial neural network (ANN) architecture is used to replace the entire code in higher level languages, running into a few thousand lines. FPGA implementation of a compact neural network for the entire range of thermodynamic properties is presented. Large arguments in sigmoid function are factored into a product of integer and a fractional part which is represented using series approximation with five terms only and the integers are represented in look up table (LUT). This ensures optimum storage and computational burden for the above applications. The ANN is implemented in IEEE 754 floating point with synthesis in Xilinx ISE design suite using Verilog HDL. The results are presented for a typical pressure versus saturation temperature.

show abstract