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.

Plasmonic nearfield scanning probe with high transmission

Abstract: Nearfield scanning optical microscopy (NSOM) offers a practical means of optical imaging, optical sensing, and nanolithography at a resolution below the diffraction limit of the light. However, its applications are limited due to the strong attenuation of the light transmitted through the subwavelength aperture. To solve this problem, we report the development of plasmonic nearfield scanning optical microscope with an efficient nearfield focusing. By exciting surface plasmons, plasmonic NSOM probes are capable of confining light into a 100 nm spot. We show by nearfield lithography experiments that the intensity at the near field is at least one order stronger than the intensity obtained from the conventional NSOM probes under the same illumination condition. Such a high efficiency can enable plasmonic NSOM as a practical tool for nearfield lithography, data storage, cellular visualization, and many other applications requiring efficient transmission with high resolution.

Fatigue and fracture behaviour of laser powder bed fusion stainless steel 316L: Influence of processing parameters

Abstract: The laser powder bed fusion (L-PBF) process involves a large number of processing parameters. Extending the intricate relationship between processing and structure to mechanical performance is essential for structural L-PBF materials. The high cycle fatigue properties of L-PBF parts are very sensitive to process-induced porosities which promote premature failure through the crack initiation mechanisms. Results from this work show that for stainless steel 316 L, porosity does not impinge on the high cycle fatigue properties when processing is kept within a ±30% tolerance band. In this ‘optimum’ processing region, crack initiation takes place due to defects at the solidification microstructure level. Beyond the ‘optimum’ processing region, over-melting and under-melting can lead to porosity-driven cracking and inferior fatigue resistance. In addition, regardless of the processing condition, fatigue resistance was found to follow a direct linear relationship with ductility and tensile strength in the low and high stress fatigue regimes respectively.

Tunable and augmented plasmon resonances of Au∕SiO2∕Au nanodisks

Abstract: The plasmon resonance of Au∕SiO2 multilayered nanodisks was studied using light scattering spectroscopy and numerical calculations. Compared to single layered Au nanodisks, multilayered nanodisks exhibit several distinctive properties including significantly enhanced plasmon resonances and tunable resonance wavelengths which can be tailored to desired values by simply varying dielectric layer thickness while the particle diameter is kept constant. Numerical calculations show that slicing one metal layer into metal multilayers leads to higher scattering intensity and more “hot spots,” or regions of strong field enhancement. This tunable and augmented plasmon resonance holds a great potential in the applications of surface-enhanced Raman scattering (SERS).

Near-field two-photon nanolithography using an apertureless optical probe

Abstract: Near-field two-photon optical lithography is demonstrated by using ∼120 fs laser pulses at 790 nm in an apertureless near-field optical microscope, which produces lithographic features with ∼70 nm resolution. The technique takes advantage of the field enhancement at the extremity of a metallic probe to induce nanoscale two-photon absorption and polymerization in a commercial photoresist, SU-8. Even without optimization of the resist or laser pulses, the spatial resolution of this technique is as high as λ/10, nearly a factor of 2 better than techniques based on far field two-photon lithography.

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.