arXiv subjects

Xiaoqiang Li

Publications and source records attributed to Xiaoqiang Li.

25 records · Page 2Linked to original sources

arXiv2021

Unsupervised Learning of Multi-level Structures for Anomaly Detection

The main difficulty in high-dimensional anomaly detection tasks is the lack of anomalous data for training. And simply collecting anomalous data from the real world, common distributions, or the boundary of normal data manifold may face the problem of missing anomaly modes. This paper first introduces a novel method to generate anomalous data by breaking up global structures while preserving local structures of normal data at multiple levels. It can efficiently expose local abnormal structures of various levels. To fully exploit the exposed multi-level abnormal structures, we propose to train multiple level-specific patch-based detectors with contrastive losses. Each detector learns to detect local abnormal structures of corresponding level at all locations and outputs patchwise anomaly scores. By aggregating the outputs of all level-specific detectors, we obtain a model that can detect all potential anomalies. The effectiveness is evaluated on MNIST, CIFAR10, and ImageNet10 dataset, where the results surpass the accuracy of state-of-the-art methods. Qualitative experiments demonstrate our model is robust that it unbiasedly detects all anomaly modes.

cs.CV↗

arXiv2018

Learning Segmentation Masks with the Independence Prior

An instance with a bad mask might make a composite image that uses it look fake. This encourages us to learn segmentation by generating realistic composite images. To achieve this, we propose a novel framework that exploits a new proposed prior called the independence prior based on Generative Adversarial Networks (GANs). The generator produces an image with multiple category-specific instance providers, a layout module and a composition module. Firstly, each provider independently outputs a category-specific instance image with a soft mask. Then the provided instances' poses are corrected by the layout module. Lastly, the composition module combines these instances into a final image. Training with adversarial loss and penalty for mask area, each provider learns a mask that is as small as possible but enough to cover a complete category-specific instance. Weakly supervised semantic segmentation methods widely use grouping cues modeling the association between image parts, which are either artificially designed or learned with costly segmentation labels or only modeled on local pairs. Unlike them, our method automatically models the dependence between any parts and learns instance segmentation. We apply our framework in two cases: (1) Foreground segmentation on category-specific images with box-level annotation. (2) Unsupervised learning of instance appearances and masks with only one image of homogeneous object cluster (HOC). We get appealing results in both tasks, which shows the independence prior is useful for instance segmentation and it is possible to unsupervisedly learn instance masks with only one image.

cs.CV↗

arXiv2018

StegNet: Mega Image Steganography Capacity with Deep Convolutional Network

Traditional image steganography often leans interests towards safely embedding hidden information into cover images with payload capacity almost neglected. This paper combines recent deep convolutional neural network methods with image-into-image steganography. It successfully hides the same size images with a decoding rate of 98.2% or bpp (bits per pixel) of 23.57 by changing only 0.76% of the cover image on average. Our method directly learns end-to-end mappings between the cover image and the embedded image and between the hidden image and the decoded image. We~further show that our embedded image, while with mega payload capacity, is still robust to statistical analysis.

cs.MM↗

arXiv2016

Monolayer MoS2/GaAs heterostructure self-driven photodetector with extremely high detectivity

Two dimensional material/semiconductor heterostructures offer alternative platforms for optoelectronic devices other than conventional Schottky and p-n junction devices. Herein, we use MoS2/GaAs heterojunction as a self-driven photodetector with wide response band width from ultraviolet to visible light, which exhibits high sensitivity to the incident light of 635 nm with responsivity as 446 mA/W and detectivity as 5.9*10^13 Jones (Jones = cm Hz1/2 W-1), respectively. Employing interface design by inserting h-BN and photo-induced doping by covering Si quantum dots on the device, the responsivity is increased to 419 mA/W for incident light of 635 nm. Distinctly, attributing to the low dark current of the MoS2/h-BN/GaAs sandwich structure based on the self-driven operation condition, the detectivity shows extremely high value of 1.9*10^14 Jones for incident light of 635 nm, which is higher than all the reported values of the MoS2 based photodetectors. Further investigations reveal that the MoS2/GaAs based photodetectors have response speed with the typical rise/fall time as 17/31 μs. The photodetectors are stable while sealed with polymethyl methacrylate after storage in air for one month. These results imply that monolayer MoS2/GaAs heterojunction may have great potential for practical applications as high performance self-driven photodetectors.

physics.ins-det↗

arXiv2015

Semiconductor quantum dots enhanced graphene/CdTe heterostructure solar cells by photo-induced doping

We report a type of solar cells based on graphene/CdTe Schottky heterostructure, which can be improved by surface engineering as graphene is one-atomic thin. By coating a layer of ultrathin CdSe quantum dots onto graphene/CdTe heterostructure, the power conversion efficiency is increased from 2.08% to 3.1%. Photo-induced doping is mainly accounted for this enhancement, as evidenced by transport, photoluminescence and quantum efficiency measurements. This work demonstrates a feasible way of designing solar cells with incorporating one dimensional and two dimensional materials.

physics.optics↗

arXiv2015

Opening the band gap of graphene through silicon doping for improved performance of graphene/GaAs heterojunction solar cells

Graphene has attracted increasing interests due to its remarkable properties, however, the zero band gap of monolayer graphene might limit its further electronic and optoelectronic applications. Herein, we have successfully synthesized monolayer silicon-doped graphene (SiG) in large area by chemical vapor deposition method. Raman spectroscopy and X-ray photoelectron spectroscopy measurements evidence silicon atoms are doped into graphene lattice with the doping level of 3.4 at%. The electrical measurement based on field effect transistor indicates that the band gap of graphene has been opened by silicon doping, which is around 0. 28 eV supported by the first-principle calculations, and the ultraviolet photoelectron spectroscopy demonstrates the work function of SiG is 0.13 eV larger than that of graphene. Moreover, the SiG/GaAs heterostructure solar cells show an improved power conversion efficiency of 33.7% in average than that of graphene/GaAs solar cells, which are attributed to the increased barrier height and improved interface quality. Our results suggest silicon doping can effectively engineer the band gap of monolayer graphene and SiG has great potential in optoelectronic device applications.

cond-mat.mtrl-sci↗

arXiv2015

High performance solar cells based on graphene-GaAs heterostructures

The honeycomb connection of carbon atoms by covalent bonds in a macroscopic two-dimensional scale leads to fascinating graphene and solar cell based on graphene/silicon Schottky diode has been widely studied. For solar cell applications, GaAs is superior to silicon as it has a direct band gap of 1.42 eV and its electron mobility is six times of that of silicon. However, graphene/GaAs solar cell has been rarely explored. Herein, we report graphene/GaAs solar cells with conversion efficiency (Eta) of 10.4% and 15.5% without and with anti-reflection layer on graphene, respectively. The Eta of 15.5% is higher than the state of art efficiency for graphene/Si system (14.5%). Furthermore, our calculation points out Eta of 25.8% can be reached by reasonably optimizing the open circuit voltage, junction ideality factor, resistance of graphene and metal/graphene contact. This research strongly support graphene/GaAs hetero-structure solar cell have great potential for practical applications.

cond-mat.mes-hall↗