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.

Coherence-Driven Topological Transition in Quantum Metamaterials.

Abstract: We introduce and theoretically demonstrate a quantum metamaterial made of dense ultracold neutral atoms loaded into an inherently defect-free artificial crystal of light, immune to well-known critical challenges inevitable in conventional solid-state platforms. We demonstrate an all-optical control, on ultrafast time scales, over the photonic topological transition of the isofrequency contour from an open to closed topology at the same frequency. This atomic lattice quantum metamaterial enables a dynamic manipulation of the decay rate branching ratio of a probe quantum emitter by more than an order of magnitude. Our proposal may lead to practically lossless, tunable, and topologically reconfigurable quantum metamaterials, for single or few-photon-level applications as varied as quantum sensing, quantum information processing, and quantum simulations using metamaterials.

Unidirectional Perfect Absorber

Abstract: We show an interplay between Fano resonances and a judicious absorption mechanism leads to a unidirectional perfect absorber, which can be controlled in both direction and frequency. Critical coupling phenomenon created by interference, separates the left- and right-side of the system. At the same time, Fano resonance causes a divergence in the delay time of photons traveling through the loss part of the system, which results in full absorption of the photons from one side. Moreover, we depict that coincidence of the two unidirectional perfect absorber modes from opposite directions results in a perfect absorber mode, which is distinct from the CPA modes. Furthermore, we show that the unidirectional perfect absorber mode is at the same time a spectral singularity and an exceptional point, which makes this point ultrasensitive to any changes in the system. Our results open a direction for designing new type of absorbers, sensors, and switches.

Synthesis of a gold nanoparticle dimer plasmonic resonator through two-phase-mediated functionalization

Abstract: We report that Au nanoparticles, ligand-exchanged with a thiol ligand at the liquid-liquid interface, were dimerized using an N,N'-diisopropylcarbodiimide-mediated amide bond formation. This dimerization of 60 nm sized Au nanoparticles achieved 24% overall yield and was visually confirmed by transmission electron microscopy as well as by scanning electron microscopy images. The resultant electromagnetic field enhancement of a single Au nanoparticle dimer was proven by dark field spectroscopy which, in turn, made the Au nanoparticle dimer suitable for molecular sensing applications, such as in surface enhanced Raman spectroscopy. Our dimerization method demonstrated that the synthesis of Au nanoparticle dimers with a high yield and enhanced optical properties of the dimers were possible. Our methodology also has good prospects as regards the formation of nanoscale building blocks.

Observation of spin-glass behavior in antiperovskite compound Mn3Cu0.7Ga0.3N

Abstract: We present a detailed study of magnetic properties of antiperovsite compound Mn3Cu0.7Ga0.3N. Ac susceptibility measurements show a peak around “freezing temperature” (Tf), with the peak position shifting as a function of the driving frequency f and magnetic field H. Magnetic relaxation measurements show a slow decay of the remanent magnetization with time below Tf. These findings consistently demonstrate the existence of spin-glass states in Mn3Cu0.7Ga0.3N. The behavior may be attributed to either a small amount of disorder, arising from the random occupation of 1a sites in the space group Pm-3m by mixed Cu/Ga atoms, or the common frustration, or both.

The Aggregation of Urban Building Clusters Based on the Skeleton Partitioning of Gap Space

Abstract: This paper presents a model for building cluster distribution analysis based on the Delaunay triangulation skeleton. The skeleton connection within the gap area among the building polygons obtains a special geometric construction similar to Voronoi diagram that spatially partitions the gap area equally. Each building polygon is surrounded by a partitioning polygon which can be regarded as the growth region of inner building. Based on this model, several cluster structure variables can be computed, such as the distribution density, the topological neighbour, the adjacent distance and the adjacent direction. Considering the constraints of position accuracy, statistical area balance, orthogonal shape in building generalization, the study presents a progressive algorithm of building cluster aggregation, including the conflict detection (where), the object (who) displacement and the geometric combination (how). The algorithm has been realized in a generalization system and some experiment illustrations are provided in the paper.

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.