Xiang Zhang

Bio: Xiang Zhang is an academic researcher from Baylor College of Medicine. The author has contributed to research in topics: Medicine & Computer science. The author has an hindex of 154, co-authored 1733 publications receiving 117576 citations. Previous affiliations of Xiang Zhang include University of California, Berkeley & University of Texas MD Anderson Cancer Center.

Resonant phase matching of Josephson junction traveling wave parametric amplifiers.

Abstract: We propose a technique to overcome phase mismatch in Josephson-junction traveling wave parametric amplifiers in order to achieve high gain over a broad bandwidth. Using ``resonant phase matching,'' we design a compact superconducting device consisting of a transmission line with subwavelength resonant inclusions that simultaneously achieves a gain of 20 dB, an instantaneous bandwidth of 3 GHz, and a saturation power of $\ensuremath{-}98\text{ }\text{ }\mathrm{dBm}$. Such an amplifier is well suited to cryogenic broadband microwave measurements such as the multiplexed readout of quantum coherent circuits based on superconducting, semiconducting, or nanomechanical elements, as well as traditional astronomical detectors.

Multiferroicity in atomic van der Waals heterostructures.

Abstract: Materials that are simultaneously ferromagnetic and ferroelectric - multiferroics - promise the control of disparate ferroic orders, leading to technological advances in microwave magnetoelectric applications and next generation of spintronics. Single-phase multiferroics are challenged by the opposite d-orbital occupations imposed by the two ferroics, and heterogeneous nanocomposite multiferroics demand ingredients' structural compatibility with the resultant multiferroicity exclusively at inter-materials boundaries. Here we propose the two-dimensional heterostructure multiferroics by stacking up atomic layers of ferromagnetic Cr2Ge2Te6 and ferroelectric In2Se3, thereby leading to all-atomic multiferroicity. Through first-principles density functional theory calculations, we find as In2Se3 reverses its polarization, the magnetism of Cr2Ge2Te6 is switched, and correspondingly In2Se3 becomes a switchable magnetic semiconductor due to proximity effect. This unprecedented multiferroic duality (i.e., switchable ferromagnet and switchable magnetic semiconductor) enables both layers for logic applications. Van der Waals heterostructure multiferroics open the door for exploring the low-dimensional magnetoelectric physics and spintronic applications based on artificial superlattices.

Electrospun TiO2–Graphene Composite Nanofibers as a Highly Durable Insertion Anode for Lithium Ion Batteries

Abstract: We report the synthesis and electrochemical performance of one-dimensional TiO2–graphene composite nanofibers (TiO2–G nanofibers) by a simple electrospinning technique for the first time. Structural and morphological properties were characterized by various techniques, such as X-ray diffraction, scanning electron microscopy (SEM), transmission electron microscopy (TEM), Raman spectroscopy, and BET surface area analysis. Lithium insertion properties were evaluated by both galvanostatic and potentiostatic modes in half-cell configurations. Cyclic voltammetric study reveals the Li-insertion/extraction by a two-phase reaction mechanism that is supported by galvanostatic charge–discharge profiles. Li/TiO2–G half-cells showed an initial discharge capacity of 260 mA h g–1 at current density of 33 mA g–1. Further, Li/TiO2–G cell retained 84% of reversible capacity after 300 cycles at a current density of 150 mA g–1, which is 25% higher than bare TiO2 nanofibers under the same test conditions. The cell also exhibi...

Enhanced sensing performance by the plasmonic analogue of electromagnetically induced transparency in active metamaterials

Abstract: The gain-assisted plasmonic analogue of electromagnetically induced transparency (EIT) in a metallic metamaterial is investigated for the purpose to enhance the sensing performance of the EIT-like plasmonic structure. The structure is composed of three bars in one unit, two of which are parallel to each other (dark quadrupolar element) but perpendicular to the third bar (bright dipolar element), The results show that, in addition to the high sensitivity to the refractive-index fluctuation of the surrounding medium, the figure of merit for such active EIT-like metamaterials can be greatly enhanced, which is attributed to the amplified narrow transparency peak.

Explosives detection in a lasing plasmon nanocavity

Abstract: Perhaps the most successful application of plasmonics to date has been in sensing, where the interaction of a nanoscale localized field with analytes leads to high-sensitivity detection in real time and in a label-free fashion. However, all previous designs have been based on passively excited surface plasmons, in which sensitivity is intrinsically limited by the low quality factors induced by metal losses. It has recently been proposed theoretically that surface plasmon sensors with active excitation (gain-enhanced) can achieve much higher sensitivities due to the amplification of the surface plasmons. Here, we experimentally demonstrate an active plasmon sensor that is free of metal losses and operating deep below the diffraction limit for visible light. Loss compensation leads to an intense and sharp lasing emission that is ultrasensitive to adsorbed molecules. We validated the efficacy of our sensor to detect explosives in air under normal conditions and have achieved a sub-part-per-billion detection limit, the lowest reported to date for plasmonic sensors with 2,4-dinitrotoluene and ammonium nitrate. The selectivity between 2,4-dinitrotoluene, ammonium nitrate and nitrobenzene is on a par with other state-of-the-art explosives detectors. Our results show that monitoring the change of the lasing intensity is a superior method than monitoring the wavelength shift, as is widely used in passive surface plasmon sensors. We therefore envisage that nanoscopic sensors that make use of plasmonic lasing could become an important tool in security screening and biomolecular diagnostics.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Going deeper with convolutions

Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.