Search

Your search keyword '"He, Qian"' showing total 322 results
Start Over Author "He, Qian" Publisher ieee
322 results on '"He, Qian"'

31. Region-Aware Face Swapping

Author

Xu, Chao, Zhang, Jiangning, Hua, Miao, He, Qian, Yi, Zili, and Liu, Yong

Subjects
FOS: Computer and information sciences, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition
Abstract

This paper presents a novel Region-Aware Face Swapping (RAFSwap) network to achieve identity-consistent harmonious high-resolution face generation in a local-global manner: \textbf{1)} Local Facial Region-Aware (FRA) branch augments local identity-relevant features by introducing the Transformer to effectively model misaligned cross-scale semantic interaction. \textbf{2)} Global Source Feature-Adaptive (SFA) branch further complements global identity-relevant cues for generating identity-consistent swapped faces. Besides, we propose a \textit{Face Mask Predictor} (FMP) module incorporated with StyleGAN2 to predict identity-relevant soft facial masks in an unsupervised manner that is more practical for generating harmonious high-resolution faces. Abundant experiments qualitatively and quantitatively demonstrate the superiority of our method for generating more identity-consistent high-resolution swapped faces over SOTA methods, \eg, obtaining 96.70 ID retrieval that outperforms SOTA MegaFS by 5.87$\uparrow$.

Published
2022

40. A Unified Memory and Hardware Security Module Based on the Adjustable Switching Window of Resistive Memory

Author

Bin Gao, Huaqiang Wu, Bohan Lin, He Qian, Yachuan Pang, and Jianshi Tang

Subjects
Hardware security module, Computer science, Random number generation, Physical unclonable function, 02 engineering and technology, 01 natural sciences, Statistical randomness, 0103 physical sciences, 0202 electrical engineering, electronic engineering, information engineering, Electrical and Electronic Engineering, PUF, 010302 applied physics, business.industry, 020208 electrical & electronic engineering, Process (computing), switching window, Electronic, Optical and Magnetic Materials, Resistive random-access memory, ComputingMilieux_MANAGEMENTOFCOMPUTINGANDINFORMATIONSYSTEMS, NIST, tri-functional module, TRNG, State (computer science), lcsh:Electrical engineering. Electronics. Nuclear engineering, business, lcsh:TK1-9971, Computer hardware, Biotechnology
Abstract

Physically unclonable function (PUF) and true random number generator (TRNG) are critical primitives to provide lightweight hardware protection. As area is tightly restricted in IoT applications, merging PUF and TRNG is a novel trend to achieve higher area efficiency. In this work, a novel tri-functional module is proposed and experimental demonstrated using RRAM for the first time. A two-phase forming process is designed to generate and store PUF ID utilizing the impact of different forming conditions on the switching window. As PUF ID is stored by the switching window, instead of a fixed resistance, PUF cells can be written to low resistance state and high resistance state as usual, and can be used as memory and TRNG. The inter-Hamming distance of the generated PUF IDs approaches 0.5, and with the majority voting readout strategy, low BER can be achieved in a wide range of temperatures from -40°C to 125°C. TRNG produces the true random numbers based on the parity of the number of pulses consumed in a write-zero process. The generated random numbers are uniform and uncorrelated after XORed, and the statistical randomness is verified by NIST SP800-22 tests.

Published
2020

46. Impact of Bottom Electrode Roughness on the Analog Switching Characteristics in Nanoscale RRAM Array

Author

Wen Sun, Wei Wu, Bin Gao, Wei Liu, Wenbin Zhang, Wang Kanwen, Jianshi Tang, Huaqiang Wu, and He Qian

Subjects
Resistive touchscreen, Materials science, Electrode roughness, business.industry, Resistive switching, Electrode, Linearity, Optoelectronics, Surface finish, business, Nanoscopic scale, Resistive random-access memory
Abstract

Computing-in-memory (CIM) architecture based on emerging resistive switching memories shows great potential to build energy-efficient hardware for artificial intelligence applications. For resistive random-access memory (RRAM), the bottom electrode (BE) roughness has been shown to impact the electrical characteristics [1] – [3] . However, previous works mainly focus on the effect of BE roughness on forming in single or a few RRAM devices with relatively large size (~μm) [1] – [3] . In this work, we study the impact of BE roughness in arrays of nanoscale RRAM with sizes down to 50 nm (applicable for 28nm technology BEOL), including retention, variation, and analog switching characteristics.

Published
2021

47. Identifying relaxation and random telegraph noises in filamentary analog RRAM for neuromorphic computing

Author

Huaqiang Wu, Jianshi Tang, Peng Yao, Bin Gao, Qi Hu, Meiran Zhao, Jiezhi Chen, Yue Xi, He Qian, Hao Zhenqi, and Yudeng Lin

Subjects
Neuromorphic engineering, Computer science, Relaxation effect, Electronic engineering, Deep neural networks, Relaxation (approximation), Resistive random-access memory
Abstract

The comprehensive investigation of relaxation effect remains difficult. For the first time, a characterization scheme is proposed and demonstrated to identify the relaxation and RTN in filamentary analog RRAM devices. This new method can help deeply understand the physical mechanism of relaxation, and pave the way for developing analog RRAM devices with excellent stability. The identification of the relaxation and RTN also helps design diversified RRAM based deep neural networks with reasonable computing accuracy.

Published
2021

48. A Compact Model of Analog RRAM Considering Temperature Coefficient for Neural Network Evaluation

Author

He Qian, Jiezhi Chen, Bin Gao, Feng Xu, Minghong Xu, Jianshi Tang, and Wei Wu

Subjects
Nonlinear system, Artificial neural network, Handwriting recognition, Computer science, Multilayer perceptron, Computer Science::Neural and Evolutionary Computation, Electronic engineering, Temperature measurement, Temperature coefficient, MNIST database, Resistive random-access memory
Abstract

Developing analog RRAM model for studying the effects of nonideal characteristic on neural network is necessary. In this work, we developed an analog RRAM compact model considering the nonideal characteristic of temperature coefficient, I-V nonlinearity, variability, and programing nonlinearity, which is well consistent with data measured from RRAM array. Then we evaluated the accuracy of MNIST task using RRAM model based multilayer perceptron. Especially, the effects of temperature coefficient on the computing task is studied.

Published
2021

49. Artificial Neuron with Spike Frequency Adaptation Based on Mott Memristor

Author

Jianshi Tang, Xinyi Li, Qiumeng Wei, Yanan Zhong, Huaqiang Wu, Bin Gao, and He Qian

Subjects
Spiking neural network, Computer science, business.industry, Pattern recognition, Biological neuron model, Spike frequency adaptation, Memristor, law.invention, Power demand, law, Artificial neuron, Artificial intelligence, business, MNIST database
Abstract

An artificial neuron with spike frequency adaptation (SFA) characteristic is demonstrated by integrating a Mott memristor with adaptation unit. By combining experimental measurements with circuit simulations, an adaptive neuron model is established and applied to a single-layer spiking neural network (SNN) trained with unsupervised spike-timing-dependent plasticity (STDP). The simulated classification accuracy on the MNIST dataset is improved from 80.21 % to 83.3% by incorporating SFA. Furthermore, the resulted uniform neuronal activities also alleviate the device requirements on endurance and uniformity $(> 4\times)$ in the hardware implementation of SNN.

Published
2021

50. Neuronal Firing Characteristics in the NbO2 based Mott Memristor

Author

Bin Gao, Xinyi Li, Huaqiang Wu, Xiaojian Zheng, He Qian, and Jianshi Tang

Subjects
Materials science, Quantitative Biology::Neurons and Cognition, Condensed matter physics, Artificial neural network, Subthreshold conduction, Relaxation (NMR), Memristor, Mott transition, law.invention, Threshold voltage, Neuromorphic engineering, law, Condensed Matter::Strongly Correlated Electrons, Natural bond orbital
Abstract

High-performance artificial neurons are critical to build bio-plausible artificial neural networks for neuromorphic computing. In this work, we report the fabrication and comprehensive characterizations of NbO 2 based Mott neuron. The device shows uniform neuronal firing behavior with a small cycle-to-cycle variation of 2.6% in the threshold voltage. Besides, a memory effect with subthreshold firing is observed in the NbO 2 Mott neuron, which is associated with the relaxation dynamics in the insulator-metal transition (IMT). Based on the Mott transition theory, this behavior can be attributed to the long-lived metallic grains in the NbO 2 layer. Such dynamic firing properties in the resistive switching process could have potential applications in neuromorphic circuits.

Published
2021

Catalog

Books, media, physical & digital resources
See catalog results