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.

Surface resonant states and superlensing in acoustic metamaterials

Abstract: We report that the negative material responses of acoustic metamaterials can lead to a plethora of surface resonant states. We determine that negative effective-mass density is the necessary condition for the existence of surface states on acoustic metamaterials. We offer the microscopic picture of these unique surface states; in addition, we find that these surface excitations enhance the transmission of evanescent pressure fields across the metamaterial. The evanescent pressure fields scattered from an object can be resonantly coupled and enhanced at the surface of the acoustic metamaterial, resulting in an image with resolution below the diffraction limit. This concept of acoustic superlens opens exciting opportunities to design acoustic metamaterials for ultrasonic imaging.

Self-amplifying autocrine actions of BDNF in axon development

Abstract: A critical step in neuronal development is the formation of axon/dendrite polarity, a process involving symmetry breaking in the newborn neuron. Local self-amplifying processes could enhance and stabilize the initial asymmetry in the distribution of axon/dendrite determinants, but the identity of these processes remains elusive. We here report that BDNF, a secreted neurotrophin essential for the survival and differentiation of many neuronal populations, serves as a self-amplifying autocrine factor in promoting axon formation in embryonic hippocampal neurons by triggering two nested positive-feedback mechanisms. First, BDNF elevates cytoplasmic cAMP and protein kinase A activity, which triggers further secretion of BDNF and membrane insertion of its receptor TrkB. Second, BDNF/TrkB signaling activates PI3-kinase that promotes anterograde transport of TrkB in the putative axon, further enhancing local BDNF/TrkB signaling. Together, these self-amplifying BDNF actions ensure stable elevation of local cAMP/protein kinase A activity that is critical for axon differentiation and growth.

MicroRNA-335 inhibits tumor reinitiation and is silenced through genetic and epigenetic mechanisms in human breast cancer

Abstract: Post-transcriptional regulators have emerged as robust effectors of metastasis and display deregulated expression through unknown mechanisms. Here, we reveal that the human microRNA-335 locus undergoes genetic deletion and epigenetic promoter hypermethylation in every metastatic derivative obtained from independent patients’ malignant cell populations. Genetic deletion of miR-335 is a common event in human breast cancer, is enriched for in breast cancer metastases, and also correlates with ovarian cancer recurrence. We furthermore identify miR-335 as a robust inhibitor of tumor reinitiation. We thus implicate the miR-335 locus on 7q32.2 as the first selective metastasis suppressor and tumor initiation suppressor locus in human breast cancer.

Large positive and negative lateral optical beam displacements due to surface plasmon resonance

Abstract: We report abnormally large positive and negative lateral optical beam shifts at a metal–air interface when the surface plasmon resonance of the metal is excited. The optimal thickness for minimal resonant reflection is identified as the critical thickness above which a negative beam displacement is observed. Experimental results show good agreement with theoretical predictions and the large observed bidirectional beam displacements also indicate the existence of forward and backward surface propagating waves at the surface plasmon resonance of the metal.

Charged-particle multiplicity measurement in proton-proton collisions at sqrt(s) = 0.9 and 2.36 TeV with ALICE at LHC

Abstract: Charged-particle production was studied in proton-proton collisions collected at the LHC with the ALICE detector at centre-of-mass energies 0.9 TeV and 2.36 TeV in the pseudorapidity range |eta| < 1.4. In the central region (|eta| < 0.5), at 0.9 TeV, we measure charged-particle pseudorapidity density dNch/deta = 3.02 +- 0.01 (stat.) +0.08 -0.05 (syst.) for inelastic interactions, and dNch/deta = 3.58 +- 0.01 (stat.) +0.12 -0.12 (syst.) for non-single-diffractive interactions. At 2.36 TeV, we find dNch/deta = 3.77 +- 0.01 (stat.) +0.25 -0.12 (syst.) for inelastic, and dNch/deta = 4.43 +- 0.01 (stat.) +0.17 -0.12 (syst.) for non-single-diffractive collisions. The relative increase in charged-particle multiplicity from the lower to higher energy is 24.7% +- 0.5% (stat.) +5.7% -2.8% (syst.) for inelastic and 23.7% +- 0.5% (stat.) +4.6% -1.1% (syst.) for non-single-diffractive interactions. This increase is consistent with that reported by the CMS collaboration for non-single-diffractive events and larger than that found by a number of commonly used models. The multiplicity distribution was measured in different pseudorapidity intervals and studied in terms of KNO variables at both energies. The results are compared to proton-antiproton data and to model predictions.

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.