Sheng Zhong

ORCID: 0000-0002-8658-3061
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Infrared Target Detection Methodologies
  • CCD and CMOS Imaging Sensors
  • Muscle activation and electromyography studies
  • Advanced Measurement and Detection Methods
  • Optical Systems and Laser Technology
  • Radio Frequency Integrated Circuit Design
  • GaN-based semiconductor devices and materials
  • Network Security and Intrusion Detection
  • Advanced Control Systems Optimization
  • Silicon Carbide Semiconductor Technologies
  • Genomics and Phylogenetic Studies
  • Cancer-related molecular mechanisms research
  • Reinforcement Learning in Robotics
  • Advanced Sensor and Energy Harvesting Materials
  • Sports Performance and Training
  • Model Reduction and Neural Networks
  • Advanced Power Amplifier Design
  • Spam and Phishing Detection
  • Data Mining Algorithms and Applications
  • Image Processing Techniques and Applications
  • Advanced Vision and Imaging
  • Advanced Image Fusion Techniques
  • Spatial Cognition and Navigation
  • Sustainable Industrial Ecology
  • Robot Manipulation and Learning

Huazhong University of Science and Technology
2001-2024

University of California, San Diego
2017-2024

Tsinghua University
2022-2024

Nanjing University of Chinese Medicine
2024

University of Michigan
2020-2022

Robotics Research (United States)
2020-2022

Shenzhen University Health Science Center
2017

Nanjing University of Science and Technology
2015

Shenzhen University
2013-2014

Hong Kong Polytechnic University
2013

In this article, a scalable knowledge-based neural network (KBNN) large-signal model of gallium nitride (GaN) high-electron-mobility transistors (HEMTs) with accurate trapping and self-heating effects characterization is developed. An improved empirical drain current proposed added to the as prior knowledge, thereby establishing model, including effect. A new equation buffer-related effect more accurately. Taking Angelov capacitance models KBNN are completed. Moreover, scaling...

10.1109/tmtt.2023.3248225 article EN IEEE Transactions on Microwave Theory and Techniques 2023-03-03

10.1109/cvpr52733.2024.01192 article EN 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024-06-16

Since the advent of software defined networks ({SDN}), there have been many attempts to outsource complex and costly local network functionality, i.e. middlebox, cloud in same way as outsourcing computation storage. The privacy issues, however, may thwart enterprises' willingness adopt this innovation since underlying configurations these middleboxes leak crucial confidential information which can be utilized by attackers. To address new problem, we use firewall an sample functionality...

10.48550/arxiv.1502.00389 preprint EN other-oa arXiv (Cornell University) 2015-01-01

A DSP/FPGA-based parallel architecture oriented to real-time image processing applications is presented. The structured with high performance DSPs interconnected by FPGA. Within FPGA a FIFO interconnection network and the specific data communication protocol are implemented, which interconnect 3 (TMS320C6414) effectively. measured performances in prototype proposed architecture, including inter-DSP system computing capacity, show transfer bandwidth (up 400 Mbytes/s) low latency as well...

10.1109/wcica.2006.1713959 article EN 2006-01-01

Behavior recognition technology can determine whether an email is a spam or ham based on behavior features instead of contents. Statistical are high degree distinct between and ham, comparatively inherent. This paper proposed model, developed Fuzzy Decision Tree filter system. The system computed Information Gain to analyze select emails. absolutely clear attributes does not always exist in real world. Email extracted from messages, processed by fuzzy processor data generalization, more...

10.1109/cit.2009.136 article EN 2009-01-01

Online behavioral advertising (OBA) has become one of the most successful models on Internet. Nevertheless, all existing OBA systems are broker-centric in billing phase, which means it is broker who exclusively determines advertisers' expenses and publishers' revenues. Consequently, a malicious may cheat their tallying ad clicks to overcharge advertisers or underpay publishers. Furthermore, as cannot justify bills, deny actual ask for refunds, publishers claim non-existing demand extra...

10.1109/icdcs.2015.38 article EN 2015-06-01

Retrieving an object from cluttered spaces such as cupboards, refrigerators, or bins requires tracking objects with limited no visual sensing. In these scenarios, contact feedback is necessary to estimate the pose of objects, yet are movable while their shapes and number may be unknown, making association contacts extremely difficult. While previous work has focused on multi-target tracking, assumptions therein prohibit using prior methods given only contact-sensing modality. Instead, this...

10.1109/lra.2022.3146915 article EN publisher-specific-oa IEEE Robotics and Automation Letters 2022-01-31

The vibration-based pavement roughness measurement method has garnered significant attention in recent years due to the development of low-cost sensors. vehicle calibration process is necessary before field effect changeable dynamics, restricting large-scale implementation measurement. This paper proposes a novel for It can quickly calibrate dynamic parameters through few controlled impact tests. model updating technique employed estimate parameters, resulting full-car numerical that...

10.1080/10298436.2023.2287688 article EN International Journal of Pavement Engineering 2023-12-22

We propose an approach to online model adaptation and control in the challenging case of hybrid discontinuous dynamics where actions may lead difficult-to-escape "trap" states, under a given controller. first learn for system without traps from randomly collected training set (since we do not know what will be encountered online). These "nominal" allow us perform tasks scenarios matches data, but when unexpected arise execution, must find way adapt our strategy continue attempting task. Our...

10.1109/lra.2021.3057789 article EN publisher-specific-oa IEEE Robotics and Automation Letters 2021-02-09

The study is aimed to investigate in vivo behaviors of the rectus femoris muscle during isometric contraction by integrating simultaneously recorded electromyography (EMG), mechanomyography (MMG), and ultrasonography (US). We developed an experimental platform for simultaneous acquisition EMG, MMG, US, as well torque, contraction. Features from multimodal signals images were then automatically extracted calibrated present time-varying characteristics behaviors. further applied local...

10.1109/iscas.2014.6865569 article EN 2022 IEEE International Symposium on Circuits and Systems (ISCAS) 2014-06-01

Anti-spam technology has been developed to the third generation technology, behavior recognition technology. There are many traditional classification models, among which, decision tree model is one most widely used, and a good intelligibility. But absolutely clear attributes does not always exist in real world. This paper proposed fuzzy based method for spam recognition. After preprocessing (data discretization, transformation compression continuous-value attributes), attribute...

10.1109/ispa.2009.60 article EN 2009-01-01

With the development of GaN HEMTs, solid state power amplifiers are hopeful to replace vacuum tube in more and areas. However, working voltage density HEMTs still very low. Transistor stacking is a promising solution ameliorate this problem. In letter, we combine unique high-voltage 0.25 μm GaN-on-SiC technology topology together build high-power-density high-gain stacked HEMTs. Our device combines two 20×120 connected serial work at 140 V. The output gain 45.6 dBm 17 dB, respectively, 3 GHz...

10.1587/elex.21.20240332 article EN IEICE Electronics Express 2024-08-05

10.52202/076061-0056 article EN 10th International Symposium on Project Management (ISPM2022) 2024-01-01

This paper presents Rummaging Using Mutual Information (RUMI), a method for online generation of robot action sequences to gather information about the pose known movable object in visually-occluded environments. Focusing on contact-rich rummaging, our approach leverages mutual between distribution and trajectory planning. From an observed partial point cloud, RUMI deduces compatible approximates it with workspace occupancy real time. Based this, we develop gain cost function reachability...

10.48550/arxiv.2408.10450 preprint EN arXiv (Cornell University) 2024-08-19

RNA-chromatin interactions play crucial roles in gene regulation and genome organization, but the interaction landscape remains poorly understood. In this study, we conducted an in-depth analysis of a previously published dataset on RNase-treated situ mapping RNA–genome interactome human embryonic stem cells. This globally profiles RNase-insensitive interactions. Our revealed that RNase treatment selectively preserved long-range while removing promiscuous resulting from local diffusion...

10.7554/elife.102554.1 preprint EN 2024-12-04

2-D Convolution is a simple mathematical operation which fundamental to many common image processing operators. Using FPGA implement the convolver can greatly reduce DSP's heavy burden in signal processing. But with limit resource small kernel. In this paper, An FIFO type line delayer presented serve as data buffer for convolution fetching operation. A finite state machine applied control reuse of multipliers and adders arrays. With these two techniques, limited be used larger kernel...

10.1117/12.750052 article EN Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2007-11-01

Transistor stacking is a promising technique to increase the working voltage and output power of amplifiers (PAs). A DC negative feedback mechanism in stacked PAs analyzed verified through measurement this paper. This ensures stability at conditions. Based on analysis mechanism, biasing procedure also Finally, an experiment performed verify our analysis. The results paper prove feasibility PA.

10.1587/elex.20.20230303 article EN IEICE Electronics Express 2023-07-31

In real-time image processing, with the improvement of resolution and frame rate camera imaging, not only requirement processing capacity is improving, but also optimization process improving. With regards to FPGA + DSP architecture system, there are three common methods overcome challenge above. The first using higher performance DSP. For example, core frequency or more cores can be used. second optimizing method, make algorithm accomplish same results spend less time. Last least,...

10.1117/12.2201126 article EN Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2015-10-08

Dynamic target tracking technology has been widely used in airborne fire control systems, warning onboard systems and battlefield surveillance tracking. The dynamic is automatic detection, identification positioning to the movement of camera. High-sensitivity CCD camera system Modern high-speed When scene that too bright ground large variation, images are easily saturated. gray collected directly related amount light enter camera, so image relies on adjustment through an optical system....

10.1109/cac.2015.7382801 article EN 2015-11-01

This paper proposed a novel outlier detection method, named l1-regularized isolation and regression (LOIRE), to examine torque-related transient patterns of in vivo muscle behavior from multimodal signals, including electromyography (EMG), mechanomyography (MMG) ultrasonography (US), during isometric contraction. Eight subjects performed ramp contraction knee up 90% the maximal voluntary contraction, EMG, MMG US were simultaneously recorded rectus femoris muscle. Five features, two root mean...

10.1109/embc.2014.6944554 article EN 2014-08-01

Intuitively, extrapolating an object’s trajectory should facilitate visual tracking. This intuition has motivated several investigations into whether humans extrapolate when tracking multiple objects. Surprisingly, they appear not to, largely ignoring motion, and implementing a low-level, spatial memory function instead. Why do to if extrapolation is beneficial? We address this question by interrogating its underlying assumption: that beneficial in the first place. Recent advances have...

10.1167/13.9.944 article EN cc-by-nc-nd Journal of Vision 2013-07-25

Most ultrasound elastographic methods utilize a purely elastic model to describe the liver mechanical properties. However, tissue that is dispersive and obtain an accurate measure of elasticity, viscoelastic property should be examined. The objective this study was investigate shear characteristics, as measured by elastography, fibrosis in rat evaluate diagnostic accuracy viscoelasticity for at different stages. Liver induced 37 rats using carbon tetrachloride (CCI4) with 6 controls. vitro...

10.1109/icics.2013.6782774 article EN 2013-12-01

A novel DSP/FPGA-based parallel architecture for real-time image processing is presented in this paper, DSPs are the main unit and FPGAs used to be logic units interface protocol, processing, display, synchronization communication portocol of DSP's reprogramming 422/485. The composed two modules: preprocessing module module, latter extendable better performance. Modules connected by LINK port, whose LVDS protocol has ability anti-jamming. And programs can updated easily 422/485 with PC's...

10.1117/12.832832 article EN Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2009-10-23
Coming Soon ...