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.

Observation of chiral phonons

Abstract: Chirality reveals symmetry breaking of the fundamental interaction of elementary particles In condensed matter, for example, the chirality of electrons governs many unconventional transport phenomena such as the quantum Hall effect Here we show that phonons can exhibit intrinsic chirality in monolayer tungsten diselenide The broken inversion symmetry of the lattice lifts the degeneracy of clockwise and counterclockwise phonon modes at the corners of the Brillouin zone We identified the phonons by the intervalley transfer of holes through hole-phonon interactions during the indirect infrared absorption, and we confirmed their chirality by the infrared circular dichroism arising from pseudoangular momentum conservation The chiral phonons are important for electron-phonon coupling in solids, phonon-driven topological states, and energy-efficient information processing

Transformational plasmon optics.

Abstract: We propose and demonstrate efficiently molding surface plasmon polaritons (SPPs) based on transformation optics. SPPs are surface modes of electromagnetic waves tightly bound at metal-dielectric interfaces, which allow us to scale optics beyond the diffraction limit. Taking advantage of transformation optics, here we show that the propagation of SPPs can be manipulated in a prescribed manner by careful control of the dielectric material properties adjacent to a metal. Since the metal properties are completely unaltered, this methodology provides a practical way for routing light at very small scales. For instance, our approach enables SPPs to travel at uneven and curved surfaces over a broad wavelength range, where SPPs would normally suffer significant scattering losses. In addition, a plasmonic 180° waveguide bend and a plasmonic Luneburg lens with simple designs are presented. The unique design flexibility of the transformational plasmon optics introduced here may open a new door to nano optics and downscaling of photonic circuits.

Accessing the exceptional points of parity-time symmetric acoustics.

Abstract: Parity-time (PT) symmetric systems experience phase transition between PT exact and broken phases at exceptional point. These PT phase transitions contribute significantly to the design of single mode lasers, coherent perfect absorbers, isolators, and diodes. However, such exceptional points are extremely difficult to access in practice because of the dispersive behaviour of most loss and gain materials required in PT symmetric systems. Here we introduce a method to systematically tame these exceptional points and control PT phases. Our experimental demonstration hinges on an active acoustic element that realizes a complex-valued potential and simultaneously controls the multiple interference in the structure. The manipulation of exceptional points offers new routes to broaden applications for PT symmetric physics in acoustics, optics, microwaves and electronics, which are essential for sensing, communication and imaging.

Large-scale chemical assembly of atomically thin transistors and circuits

Abstract: Next-generation electronics calls for new materials beyond silicon, aiming at increased functionality, performance and scaling in integrated circuits. In this respect, two-dimensional gapless graphene and semiconducting transition-metal dichalcogenides have emerged as promising candidates due to their atomic thickness and chemical stability. However, difficulties with precise spatial control during their assembly currently impede actual integration into devices. Here, we report on the large-scale, spatially controlled synthesis of heterostructures made of single-layer semiconducting molybdenum disulfide contacting conductive graphene. Transmission electron microscopy studies reveal that the single-layer molybdenum disulfide nucleates at the graphene edges. We demonstrate that such chemically assembled atomic transistors exhibit high transconductance (10 µS), on–off ratio (∼106) and mobility (∼17 cm2 V−1 s−1). The precise site selectivity from atomically thin conducting and semiconducting crystals enables us to exploit these heterostructures to assemble two-dimensional logic circuits, such as an NMOS inverter with high voltage gain (up to 70). Large-scale electronic circuits can be assembled via the spatially controlled synthesis of heterostructures made of single-layer molybdenum disulfide contacting graphene.

All-angle negative refraction and imaging in a bulk medium made of metallic nanowires in the visible region

Abstract: We theoretically demonstrated that all-angle negative refraction and imaging can be implemented by metallic nanowires embedded in a dielectric matrix. When the separation between the nanowires is much smaller than the incident wavelength, these structures can be characterized as indefinite media, whose effective permittivities perpendicular and parallel to the wires are opposite in signs. Under this condition, the dispersion diagram is hyperbolic for transverse magnetic waves propagating in the nanowire system, thereby exhibiting all-angle negative refraction. Such indefinite media can operate over a broad frequency range (visible to near-infrared) far from any resonances, thus they offer an effective way to manipulate light propagation in bulk media with low losses, allowing potential applications in photonic devices.

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.