National University of Defense Technology

About: National University of Defense Technology is a education organization based out in Changsha, China. It is known for research contribution in the topics: Computer science & Radar. The organization has 39430 authors who have published 40181 publications receiving 358979 citations. The organization is also known as: Guófáng Kēxuéjìshù Dàxué & NUDT.

Eliminating Negative‐SET Behavior by Suppressing Nanofilament Overgrowth in Cation‐Based Memory

Abstract: Negative-SET behavior is observed in various cation-based memories, which degrades the device reliability. Transmission electron microscopy results demonstrate the behavior is caused by the overgrowth of the conductive filament (CF) into the Pt electrode. The CF overgrowth phenomenon is suppressed and the negative-SET behavior is eliminated by inserting an impermeable graphene layer. The graphene-based devices show high reliability and satisfying performance.

ThunderNet: Towards Real-Time Generic Object Detection on Mobile Devices

Abstract: Real-time generic object detection on mobile platforms is a crucial but challenging computer vision task. Prior lightweight CNN-based detectors are inclined to use one-stage pipeline. In this paper, we investigate the effectiveness of two-stage detectors in real-time generic detection and propose a lightweight two-stage detector named ThunderNet. In the backbone part, we analyze the drawbacks in previous lightweight backbones and present a lightweight backbone designed for object detection. In the detection part, we exploit an extremely efficient RPN and detection head design. To generate more discriminative feature representation, we design two efficient architecture blocks, Context Enhancement Module and Spatial Attention Module. At last, we investigate the balance between the input resolution, the backbone, and the detection head. Benefit from the highly efficient backbone and detection part design, ThunderNet surpasses previous lightweight one-stage detectors with only 40% of the computational cost on PASCAL VOC and COCO benchmarks. Without bells and whistles, ThunderNet runs at 24.1 fps on an ARM-based device with 19.2 AP on COCO. To the best of our knowledge, this is the first real-time detector reported on ARM platforms. Code will be released for paper reproduction.

Giant electrical modulation of magnetization in Co40Fe40B20/Pb(Mg1/3Nb2/3)0.7Ti0.3O3(011) heterostructure.

Abstract: We report a giant electric-field control of magnetization (M) as well as magnetic anisotropy in a Co40Fe40B20(CoFeB)/Pb(Mg1/3Nb2/3)0.7Ti0.3O3(PMN-PT) structure at room temperature, in which a maximum relative magnetization change (ΔM/M) up to 83% with a 90° rotation of the easy axis under electric fields were observed by different magnetic measurement systems with in-situ electric fields. The mechanism for this giant magnetoelectric (ME) coupling can be understood as the combination of the ultra-high value of anisotropic in-plane piezoelectric coefficients of (011)-cut PMN-PT due to ferroelectric polarization reorientation and the perfect soft ferromagnetism without magnetocrystalline anisotropy of CoFeB film. Besides the giant electric-field control of magnetization and magnetic anisotropy, this work has also demonstrated the feasibility of reversible and deterministic magnetization reversal controlled by pulsed electric fields with the assistance of a weak magnetic field, which is important for realizing strain-mediated magnetoelectric random access memories.

Magnetic SERS Strip for Sensitive and Simultaneous Detection of Respiratory Viruses

Abstract: Rapid and early diagnosis of respiratory viruses is key to preventing infections from spreading and guiding treatments. Here, we developed a sensitive and quantitative surface-enhanced Raman scatte...

Supersonic flow imaging via nanoparticles

Abstract: Due to influence of compressibility, shock wave, instabilities, and turbulence on supersonic flows, current flow visualization and imaging techniques encounter some problems in high spatiotemporal resolution and high signal-to-noise ratio (SNR) measurements. Therefore, nanoparticle based planar laser scattering method (NPLS) is developed here. The nanoparticles are used as tracer, and pulse planar laser is used as light source in NPLS; by recording images of particles in flow field with CCD, high spatiotemporal resolution supersonic flow imaging is realized. The flow-following ability of nanoparticles in supersonic flows is studied according to multiphase flow theory and calibrating experiment of oblique shock wave. The laser scattering characteristics of nanoparticles are analyzed with light scattering theory. The results of theoretical and experimental studies show that the dynamic behavior and light scattering characteristics of nanoparticles highly enhance the spatiotemporal resolution and SNR of NPLS, with which the flow field involving shock wave, expansion, Mach disk, boundary layer, sliding-line, and mixing layer can be imaged clearly at high spatiotemporal resolution.