scispace - formally typeset
Search or ask a question
Author

Guoqiang He

Bio: Guoqiang He is an academic researcher. The author has contributed to research in topics: Computer science & Artificial neural network. The author has an hindex of 1, co-authored 1 publications receiving 13 citations.

Papers
More filters
Proceedings ArticleDOI
Bin Wang1, Bin Yang1, Jinfang Sheng1, Mengsheng Chen, Guoqiang He 
23 Jan 2009
TL;DR: The proposed conjugate gradient algorithm has global convergence, optimizes the learning steps using new line search rules and improves the convergence speed, and the new algorithm is applied in the cost prediction of actual sintering production.
Abstract: This paper studies various training algorithms of BP neural network and proposes an improved conjugate gradient algorithm which combines conjugate gradient algorithm with inexact line search route based on generalized Curry principle. The proposed algorithm has global convergence, optimizes the learning steps using new line search rules and improves the convergence speed. The new algorithm is applied in the cost prediction of actual sintering production. Simulation results show that the algorithm has better convergence compared with traditional conjugate gradient algorithms. The MSE of prediction is 0.0098 and accuracy rate reaches 94.31%.

14 citations

Proceedings ArticleDOI
28 May 2022
TL;DR: In this article , the authors proposed a low-cost spike-time based unsupervised learning method, which constructs a SNN with one fully-connected excitatory layer structure without inhibitory layer, and trains the SNN using a first-spike-based temporal coding scheme where input information is directly encoded into spike times.
Abstract: Spiking Neural Networks (SNNs) have been recognized as one of the next generation of Neural Networks (NNs), showing a great potential in a variety of applications. Spiking-Timing Dependent Plasticity (STDP) underlies the brain’s learning mechanisms, and trains SNNs with great energy efficiency. In this paper, we propose a low-cost spike-time based unsupervised learning method. It constructs a SNN with one fully-connected excitatory layer structure without inhibitory layer, and trains the SNN with STDP using a first-spike-based temporal coding scheme where input information is directly encoded into spike times. It only updates the synaptic weights connected to the neuron that first generates a spike in a forward propagation step, which reduces the frequency of the synaptic weight updates significantly. The forward propagation process can be stopped once a neuron fires whether in the training mode or the inference mode, by which many unnecessary computations are just avoided and the latency in the inference mode is reduced. The method was used to train on the classification task on MNIST dataset and achieved an accuracy of 90.4% with 800 excitatory neurons.

2 citations

Journal ArticleDOI
TL;DR: The proposed architecture uses the pretreatment for normalization and parabolic synthesis method to calculate the Nth root of modulus of the input complex number and performs the conversion between the plane coordinate form and the polar coordinate form of the complex number by CORDIC, which not only ensures the accuracy but also has an ultra-low computation latency.
Abstract: This paper presents a new architecture, based on CORDIC and parabolic synthesis methodology, for computing Nth root of a complex number. The proposed architecture uses the pretreatment for normalization and parabolic synthesis method to calculate the Nth root of modulus of the input complex number and performs the conversion between the plane coordinate form and the polar coordinate form of the complex number by CORDIC, which not only ensures the accuracy but also has an ultra-low computation latency. MATLAB simulation result indicates that our proposed method can calculate the Nth root of the complex numbers in the form of fixed-point number with an error of $2.16 \boldsymbol {\times {10^{ - 6}}}$ . Under TSMC 40nm CMOS technology, the report shows that the area consumption is $27390.72 \boldsymbol {\mu m^{2}}$ at the frequency of 1GHz and the power consumption is 2.3549mW. More importantly, the computation latency of the proposed architecture is only 60.18% of the latest architecture in the same calculation accuracy.

2 citations

Journal ArticleDOI
Guoqiang He, Xiaoling Ding, Ming Zhou, Bo Liu, Li Li 
TL;DR: This paper proposes a background noise adaptive energy-efficient keywords recognition processor with Reusable DNN (RDNN) and reconfigurable architecture and the SNR prediction module determines whether the computing mode is low power consumption mode (LPM) or high performance mode (HPM).
Abstract: This paper proposes a background noise adaptive energy-efficient keywords recognition processor with Reusable DNN (RDNN) and reconfigurable architecture. To reduce power consumption while maintaining the recognition accuracy of different background noises, the SNR prediction module determines whether the computing mode is low power consumption mode (LPM) or high performance mode (HPM). In LPM, DNN-shift (shift-based deep neural network) is used to achieve high recognition accuracy in a low background noise environment; in HPM, DNN-8bit (8bit weighted deep neural network) is used to achieve low power consumption in a high background noise environment. And the two modes share most of the hardware, and approximate computing is introduced to further reduce power consumption. Evaluated under 22nm process technology, this work can support up to 10 keywords recognition with the power consumption of 11.2μW for high background noise and 7.3μW for low background noise.

2 citations

Journal ArticleDOI
TL;DR: In this article , a DSP-purposed REconfigurable Acceleration Machine (DREAM) core is proposed for wireless baseband digital signal processing, which has a good trade-off between flexibility and performance.
Abstract: The wireless baseband processing algorithms are still developing and show a great diversity. The development of ASIC implementations cannot quickly adapt to the evolution of algorithms and standards. Meanwhile, the general-purpose processors cannot meet the real-time requirements in some scenarios. This paper proposes a DSP-purposed REconfigurable Acceleration Machine (DREAM) core for wireless baseband digital signal processing, which has a good trade-off between flexibility and performance. First, we abstract a set of shared operators with a moderate granularity from a variety of wireless MIMO signal processing algorithms. Then, we propose a two-step configuration process to reduce the size of the required reconfiguration bits. Besides, we design a conflict-free address generator to transfer data between the on-chip scratchpad memory and reconfiguration processing elements with high efficiency and high throughput. Finally, the prototype DREAM core has been implemented in TSMC CMOS 28 nm, and its area and power consumption have been analyzed. The chip has great flexibility in supporting a variety of wireless MIMO processing algorithms and a wide range of MIMO scales. The proposed DREAM core can achieve the normalized area efficiency and the normalized energy efficiency of $0.67~Gbps/MGE$ and $15.05~Gbps/W$ , which are $1.56\times $ and $4.18\times $ those of state-of-the-art reconfigurable implementations when running the WeJi-based MIMO detection algorithm.

1 citations


Cited by
More filters
Journal ArticleDOI
Min Wu1, Xiaoxia Chen1, Weihua Cao1, Jinhua She1, Chunsheng Wang1 
TL;DR: In this paper, an intelligent integrated optimization system (IIOS) was developed for the proportioning step, which contains two phases: the first and second proportionings, and the predicted quality indices were fed back to the optimizations of the first two proportionings to find feasible optimal dosing schemes.

37 citations

Journal ArticleDOI
TL;DR: In this paper, an integrated neural-network-based model for predicting the burn-through point (BTP) of a lead-zinc sintering process was presented.

21 citations

Journal ArticleDOI
TL;DR: A predictive system for two important sinter chemical composition (TFe and FeO), sinter output and sintering solid fuel consumption of was established based on BP neural network, which was trained by actual production data as discussed by the authors.
Abstract: The sinter quality and the stability of composition could directly affect the yield, quality and energy consumption of ironmaking production. It is important for iron and steel industry to steadily control sinter chemical composition and analyze sintering energy consumption. The MATLAB m file editor was used to write code directly in this paper. A predictive system for two important sinter chemical composition (TFe and FeO), sinter output and sintering solid fuel consumption of was established based on BP neural network, which was trained by actual production data.) The application results show that the prediction system has high accuracy rate, stability and reliability, the sintering productivity was improved effectively.

4 citations

Journal ArticleDOI
TL;DR: It was shown that the trained feed-forward backpropagation neural network was able to detect which IGBT of the single phase square wave inverter produced the incipient faults.
Abstract: With the increasing usage of solar Photovoltaic (PV) system in Malaysia, the condition aspect of solar PV system especially inverter system needs to be given full attention. Detection of faults at earlier stage is very important in order to avoid the extended period of down-time caused by inverter failure. Thus, this paper aims to detect the source of the incipient faults produced by a single phase inverter of a PV system. The incipient faults were generated by modifying the pulse wave control signal. A total of 100 incipient faults and one set of normal condition waveform are collected at the output of the single phase square wave inverter. These waveforms are then used to train the feed-forward backpropagation neural network. One hidden layer feed-forward backpropagation neural network of 9 neurons was trained and MSE of 6.13 × 10-4 was obtained. It was shown that the trained feed-forward backpropagation neural network was able to detect which IGBT of the single phase square wave inverter produced the incipient faults.

3 citations