arXiv subjects

Qing Li

Publications and source records attributed to Qing Li.

At least 559 records · Page 31Linked to original sources

arXiv2019

MMED: A Multi-domain and Multi-modality Event Dataset

In this work, we construct and release a multi-domain and multi-modality event dataset (MMED), containing 25,165 textual news articles collected from hundreds of news media sites (e.g., Yahoo News, Google News, CNN News.) and 76,516 image posts shared on Flickr social media, which are annotated according to 412 real-world events. The dataset is collected to explore the problem of organizing heterogeneous data contributed by professionals and amateurs in different data domains, and the problem of transferring event knowledge obtained from one data domain to heterogeneous data domain, thus summarizing the data with different contributors. We hope that the release of the MMED dataset can stimulate innovate research on related challenging problems, such as event discovery, cross-modal (event) retrieval, and visual question answering, etc.

cs.MM↗

arXiv2019

Ultrashort Echo Time Magnetic Resonance Fingerprinting (UTE-MRF) for Simultaneous Quantification of Long and Ultrashort T2 Tissues

Purpose: To demonstrate an ultrashort echo time magnetic resonance fingerprinting (UTE-MRF) method that can simultaneously quantify tissue relaxometries for muscle and bone in musculoskeletal systems and tissue components in brain and therefore can synthesize pseudo-CT images. Methods: A FISP-MRF sequence with half pulse excitation and half spoke radial acquisition was designed to sample fast T2 decay signals. Sinusoidal echo time (TE) pattern was applied to enhance MRF sensitivity for tissues with short and ultrashort T2 values. The performance of UTE-MRF was evaluated via simulations, phantoms, and in vivo experiments. Results: A minimal TE of 0.05 ms was achieved in UTE-MRF. Simulations indicated that extension of TE sampling increased T2 quantification accuracy in cortical bone and tendon, and had little impact on long T2 muscle quantifications. For a rubber phantom, an average T1/T2 of 162/1.07 ms from UTE-MRF were compared well with gold standard T2 of 190 ms from IR-UTE and T2* of 1.03 ms from UTE sequence. For a long T2 agarose phantom, the linear regression slope between UTE-MRF and gold standard was 1.07 (R2=0.991) for T1 and 1.04 (R2=0.994) for T2. In vivo experiments showed the detection of cortical bone and Achilles tendon, where the averaged T2 was respectively 1.0 ms and 15 ms. Scalp images were in good agreement with CT. Conclusion: UTE-MRF with sinusoidal TE variations shows its capability to produce pseudo-CT images and simultaneously output T1, T2, proton density, and B0 maps for tissues with long T2 and short/ultrashort T2 in the brain and musculoskeletal system.

physics.med-ph↗

arXiv2019

A Hybrid Data Cleaning Framework using Markov Logic Networks

With the increase of dirty data, data cleaning turns into a crux of data analysis. Most of the existing algorithms rely on either qualitative techniques (e.g., data rules) or quantitative ones (e.g., statistical methods). In this paper, we present a novel hybrid data cleaning framework on top of Markov logic networks (MLNs), termed as MLNClean, which is capable of cleaning both schema-level and instance-level errors. MLNClean mainly consists of two cleaning stages, namely, first cleaning multiple data versions separately (each of which corresponds to one data rule), and then deriving the final clean data based on multiple data versions. Moreover, we propose a series of techniques/concepts, e.g., the MLN index, the concepts of reliability score and fusion score, to facilitate the cleaning process. Extensive experimental results on both real and synthetic datasets demonstrate the superiority of MLNClean to the state-of-the-art approach in terms of both accuracy and efficiency.

cs.DB↗

arXiv2019

Efficient telecom-to-visible spectral translation through ultra-low power nonlinear nanophotonics

The ability to spectrally translate lightwave signals in a compact, low-power platform is at the heart of the promise of nonlinear nanophotonic technologies. For example, a device to link the telecommunications band with visible and short near-infrared wavelengths can enable a connection between high-performance chip-integrated lasers based on scalable nanofabrication technology with atomic systems used for time and frequency metrology. While second-order nonlinear (χ^(2)) systems are the natural approach for bridging such large spectral gaps, here we show that third-order nonlinear (chi^(3)) systems, despite their typically much weaker nonlinear response, can realize spectral translation with unprecedented performance. By combining resonant enhancement with nanophotonic mode engineering in a silicon nitride microring resonator, we demonstrate efficient spectral translation of a continuous-wave signal from the telecom band (~ 1550 nm) to the visible band (~ 650 nm) through cavity-enhanced four-wave mixing. We achieve such translation over a wide spectral range >250 THz with a translation efficiency of (30.1 +/- 2.8) % and using an ultra-low pump power of (329 +/- 13) uW. The translation efficiency projects to (274 +/- 28) % at 1 mW and is more than an order of magnitude larger than what has been achieved in current nanophotonic devices.

physics.optics↗

arXiv2019

A Roadmap-Path Reshaping Algorithm for Real-Time Motion Planning

Real-time motion planning is a vital function of robotic systems. Different from existing roadmap algorithms which first determine the free space and then determine the collision-free path, researchers recently proposed several convex relaxation based smoothing algorithms which first select an initial path to link the starting configuration and the goal configuration and then reshape this path to meet other requirements (e.g., collision-free conditions) by using convex relaxation. However, convex relaxation based smoothing algorithms often fail to give a satisfactory path, since the initial paths are selected randomly. Moreover, the curvature constraints were not considered in the existing convex relaxation based smoothing algorithms. In this paper, we show that we can first grid the whole configuration space to pick a candidate path and reshape this shortest path to meet our goal. This new algorithm inherits the merits of the roadmap algorithms and the convex feasible set algorithm. We further discuss how to meet the curvature constraints by using both the Beamlet algorithm to select a better initial path and an iterative optimization algorithm to adjust the curvature of the path. Theoretical analyzing and numerical testing results show that it can almost surely find a feasible path and use much less time than the recently proposed convex feasible set algorithm.

cs.RO↗

arXiv2019

Enhancing Remote Sensing Image Retrieval with Triplet Deep Metric Learning Network

With the rapid growing of remotely sensed imagery data, there is a high demand for effective and efficient image retrieval tools to manage and exploit such data. In this letter, we present a novel content-based remote sensing image retrieval method based on Triplet deep metric learning convolutional neural network (CNN). By constructing a Triplet network with metric learning objective function, we extract the representative features of the images in a semantic space in which images from the same class are close to each other while those from different classes are far apart. In such a semantic space, simple metric measures such as Euclidean distance can be used directly to compare the similarity of images and effectively retrieve images of the same class. We also investigate a supervised and an unsupervised learning methods for reducing the dimensionality of the learned semantic features. We present comprehensive experimental results on two publicly available remote sensing image retrieval datasets and show that our method significantly outperforms state-of-the-art.

cs.CV↗

arXiv2019

Relative Geometry-Aware Siamese Neural Network for 6DOF Camera Relocalization

6DOF camera relocalization is an important component of autonomous driving and navigation. Deep learning has recently emerged as a promising technique to tackle this problem. In this paper, we present a novel relative geometry-aware Siamese neural network to enhance the performance of deep learning-based methods through explicitly exploiting the relative geometry constraints between images. We perform multi-task learning and predict the absolute and relative poses simultaneously. We regularize the shared-weight twin networks in both the pose and feature domains to ensure that the estimated poses are globally as well as locally correct. We employ metric learning and design a novel adaptive metric distance loss to learn a feature that is capable of distinguishing poses of visually similar images from different locations. We evaluate the proposed method on public indoor and outdoor benchmarks and the experimental results demonstrate that our method can significantly improve localization performance. Furthermore, extensive ablation evaluations are conducted to demonstrate the effectiveness of different terms of the loss function.

cs.CV↗

arXiv2018

The effect of oscillator and dipole-dipole interaction on multiple optomechanically induced transparency in cavity optomechanical system

We theoretically investigate the optomechanically induced transparency (OMIT) phenomenon in a N-cavity optomechanical system doped with a pair of Rydberg atoms with the presence of a strong pump field and a weak probe field applied to the Nth cavity. 2N-1(N<10) number OMIT windows can be observed in the output field when N cavities coupled with N mechanical oscillators, respectively. But, the mechanical oscillators coupled with different even-odd label cavities lead to different effect on OMIT. On the other hand, two additional transparent windows (extra resonances) are presented, if two Rydberg atoms are coupled with the cavity field. With the DDI increasing, it is interesting that the extra resonances move to right and the left extra resonance moves slowly than the right one. During this process, Fano resonance is also shown on the output field.

quant-ph↗

arXiv2018

A Kerr-microresonator optical clockwork

Kerr microresonators generate interesting and useful fundamental states of electromagnetic radiation through nonlinear interactions of continuous-wave (CW) laser light. Using photonic-integration techniques, functional devices with low noise, small size, low-power consumption, scalable fabrication, and heterogeneous combinations of photonics and electronics can be realized. Kerr solitons, which stably circulate in a Kerr microresonator, have emerged as a source of coherent, ultrafast pulse trains and ultra-broadband optical-frequency combs. Using the f-2f technique, Kerr combs support carrier-envelope-offset phase stabilization for optical synthesis and metrology. In this paper, we introduce a Kerr-microresonator optical clockwork based on optical-frequency division (OFD), which is a powerful technique to transfer the fractional-frequency stability of an optical clock to a lower frequency electronic clock signal. The clockwork presented here is based on a silicon-nitride (Si$_3$N$_4$) microresonator that supports an optical-frequency comb composed of soliton pulses at 1 THz repetition rate. By electro-optic phase modulation of the entire Si$_3$N$_4$ comb, we arbitrarily generate additional CW modes between the Si$_3$N$_4$ comb modes; operationally, this reduces the pulse train repetition frequency and can be used to implement OFD to the microwave domain. Our experiments characterize the residual frequency noise of this Kerr-microresonator clockwork to one part in $10^{17}$, which opens the possibility of using Kerr combs with high performance optical clocks. In addition, the photonic integration and 1 THz resolution of the Si$_3$N$_4$ frequency comb makes it appealing for broadband, low-resolution liquid-phase absorption spectroscopy, which we demonstrate with near infrared measurements of water, lipids, and organic solvents.

physics.optics↗

arXiv2018

The healing strategy by prioritizing minimum degree against localized attacks on interdependent spatially embedded networks

Many real infrastructure networks, such as power grids and communication networks, are not only depend on one another to function, but also embedded in space. A lot of works have been devoted to reveal the vulnerability of interdependent spatially embedded networks considers random failures. However, recently research show that they are susceptible to geographically localized attacks or failures cased by natural disasters or terrorist attacks, which affect all nodes within a given radius. In particular, small localized attacks may lead to catastrophic consequences. As a remedy of the collapse instability, one research introduced a dynamic healing strategy and the possibility of new connectivity link with a probability, called random healing, to bridge two remaining neighbors of a failed node. The random strategy is straightforward. Here, unlike previous strategy, we propose a simple but effective strategy, called healing strategy by prioritizing minimum degree(HPMD), which establishing a new link between two functioning neighbors with minimum degree during the cascading failures. In addition, we consider the distance between two nodes to avoid change the network structures. Afterwards, a comparison is made between HPMD and three healing strategies: random, degree centrality and local centrality to identify which a pair of neighbors have high priority. Simulation experiments are presented for two square lattices with the same size as interdependent spatially embedded networks under localized attacks. Results demonstrated that HPMD strategy has an outstanding performance against localized attacks, even when ratio of healing is vary. Moreover, HPMD is more timely, more applicability and costless. Our method is meaningful in practice as it can greatly enhance the resilience and robustness of interdependent spatially embedded networks against localized attacks.

physics.soc-ph↗

arXiv2018

On the Effectiveness of Least Squares Generative Adversarial Networks

Unsupervised learning with generative adversarial networks (GANs) has proven to be hugely successful. Regular GANs hypothesize the discriminator as a classifier with the sigmoid cross entropy loss function. However, we found that this loss function may lead to the vanishing gradients problem during the learning process. To overcome such a problem, we propose in this paper the Least Squares Generative Adversarial Networks (LSGANs) which adopt the least squares loss for both the discriminator and the generator. We show that minimizing the objective function of LSGAN yields minimizing the Pearson $χ^2$ divergence. We also show that the derived objective function that yields minimizing the Pearson $χ^2$ divergence performs better than the classical one of using least squares for classification. There are two benefits of LSGANs over regular GANs. First, LSGANs are able to generate higher quality images than regular GANs. Second, LSGANs perform more stably during the learning process. For evaluating the image quality, we conduct both qualitative and quantitative experiments, and the experimental results show that LSGANs can generate higher quality images than regular GANs. Furthermore, we evaluate the stability of LSGANs in two groups. One is to compare between LSGANs and regular GANs without gradient penalty. We conduct three experiments, including Gaussian mixture distribution, difficult architectures, and a newly proposed method --- datasets with small variability, to illustrate the stability of LSGANs. The other one is to compare between LSGANs with gradient penalty (LSGANs-GP) and WGANs with gradient penalty (WGANs-GP). The experimental results show that LSGANs-GP succeed in training for all the difficult architectures used in WGANs-GP, including 101-layer ResNet.

cs.CV↗

arXiv2018

VQA-E: Explaining, Elaborating, and Enhancing Your Answers for Visual Questions

Most existing works in visual question answering (VQA) are dedicated to improving the accuracy of predicted answers, while disregarding the explanations. We argue that the explanation for an answer is of the same or even more importance compared with the answer itself, since it makes the question and answering process more understandable and traceable. To this end, we propose a new task of VQA-E (VQA with Explanation), where the computational models are required to generate an explanation with the predicted answer. We first construct a new dataset, and then frame the VQA-E problem in a multi-task learning architecture. Our VQA-E dataset is automatically derived from the VQA v2 dataset by intelligently exploiting the available captions. We have conducted a user study to validate the quality of explanations synthesized by our method. We quantitatively show that the additional supervision from explanations can not only produce insightful textual sentences to justify the answers, but also improve the performance of answer prediction. Our model outperforms the state-of-the-art methods by a clear margin on the VQA v2 dataset.

cs.CV↗

arXiv2018

Single Bitmap Block Truncation Coding of Color Images Using Hill Climbing Algorithm

Recently, the use of digital images in various fields is increasing rapidly. To increase the number of images stored and get faster transmission of them, it is necessary to reduce the size of these images. Single bitmap block truncation coding (SBBTC) schemes are compression techniques, which are used to generate a common bitmap to quantize the R, G and B planes in color image. As one of the traditional SBBTC schemes, weighted plane (W-plane) method is famous for its simplicity and low time consumption. However, the W-plane method also has poor performance in visual quality. This paper proposes an improved SBBTC scheme based on W-plane method using parallel computing and hill climbing algorithm. Compared with various schemes, the simulation results of the proposed scheme are better than that of the reference schemes in visual quality and time consumption.

cs.CV↗

arXiv2018

Chip-integrated visible-telecom photon pair sources for quantum communication

Photon pair sources are fundamental building blocks for quantum entanglement and quantum communication. Recent studies in silicon photonics have documented promising characteristics for photon pair sources within the telecommunications band, including sub-milliwatt optical pump power, high spectral brightness, and high photon purity. However, most quantum systems suitable for local operations, such as storage and computation, support optical transitions in the visible or short near-infrared bands. In comparison to telecommunications wavelengths, the significantly higher optical attenuation in silica at such wavelengths limits the length scale over which optical-fiber-based quantum communication between such local nodes can take place. One approach to connect such systems over fiber is through a photon pair source that can bridge the visible and telecom bands, but an appropriate source, which should produce narrow-band photon pairs with a high signal-to-noise ratio, has not yet been developed in an integrated platform. Here, we demonstrate a nanophotonic visible-telecom photon pair source for the first time, using high quality factor silicon nitride resonators to generate bright photon pairs with an unprecedented coincidence-to-accidental ratio (CAR) up to (3800 +/- 200). We further demonstrate dispersion engineering of the microresonators to enable the connection of different species of trapped atoms/ions, defect centers, and quantum dots to the telecommunications bands for future quantum communication systems.

physics.optics↗

arXiv2018

VizWiz Grand Challenge: Answering Visual Questions from Blind People

The study of algorithms to automatically answer visual questions currently is motivated by visual question answering (VQA) datasets constructed in artificial VQA settings. We propose VizWiz, the first goal-oriented VQA dataset arising from a natural VQA setting. VizWiz consists of over 31,000 visual questions originating from blind people who each took a picture using a mobile phone and recorded a spoken question about it, together with 10 crowdsourced answers per visual question. VizWiz differs from the many existing VQA datasets because (1) images are captured by blind photographers and so are often poor quality, (2) questions are spoken and so are more conversational, and (3) often visual questions cannot be answered. Evaluation of modern algorithms for answering visual questions and deciding if a visual question is answerable reveals that VizWiz is a challenging dataset. We introduce this dataset to encourage a larger community to develop more generalized algorithms that can assist blind people.

cs.CV↗

arXiv2018

Unpaired Multi-Domain Image Generation via Regularized Conditional GANs

In this paper, we study the problem of multi-domain image generation, the goal of which is to generate pairs of corresponding images from different domains. With the recent development in generative models, image generation has achieved great progress and has been applied to various computer vision tasks. However, multi-domain image generation may not achieve the desired performance due to the difficulty of learning the correspondence of different domain images, especially when the information of paired samples is not given. To tackle this problem, we propose Regularized Conditional GAN (RegCGAN) which is capable of learning to generate corresponding images in the absence of paired training data. RegCGAN is based on the conditional GAN, and we introduce two regularizers to guide the model to learn the corresponding semantics of different domains. We evaluate the proposed model on several tasks for which paired training data is not given, including the generation of edges and photos, the generation of faces with different attributes, etc. The experimental results show that our model can successfully generate corresponding images for all these tasks, while outperforms the baseline methods. We also introduce an approach of applying RegCGAN to unsupervised domain adaptation.

cs.CV↗

arXiv2018

Phased-Locked Two-Color Single Soliton Microcombs in Dispersion-Engineered Si3N4 Resonators

We propose and theoretically investigate a dispersion-engineered Si3N4 microring resonator, based on a cross-section containing a partially-etched trench, that supports phase-locked, two-color soliton microcomb states. These soliton states consist of a single circulating intracavity pulse with a modulated envelope that sits on a continuous wave background. Such temporal waveforms produce a frequency comb whose spectrum is spread over two widely-spaced spectral windows, each exhibiting a squared hyperbolic secant envelope, with the two windows phase-locked to each other via Cherenkov radiation. The first spectral window is centered around the 1550 nm pump, while the second spectral window is tailored based on straightforward geometric control, and can be centered as short as 750 nm and as long as 3000 nm. We numerically analyze the robustness of the design to parameter variation, and consider its implications to self-referencing and visible wavelength comb generation.

physics.optics↗

arXiv2018

Unsupervised Cross-dataset Person Re-identification by Transfer Learning of Spatial-Temporal Patterns

Most of the proposed person re-identification algorithms conduct supervised training and testing on single labeled datasets with small size, so directly deploying these trained models to a large-scale real-world camera network may lead to poor performance due to underfitting. It is challenging to incrementally optimize the models by using the abundant unlabeled data collected from the target domain. To address this challenge, we propose an unsupervised incremental learning algorithm, TFusion, which is aided by the transfer learning of the pedestrians' spatio-temporal patterns in the target domain. Specifically, the algorithm firstly transfers the visual classifier trained from small labeled source dataset to the unlabeled target dataset so as to learn the pedestrians' spatial-temporal patterns. Secondly, a Bayesian fusion model is proposed to combine the learned spatio-temporal patterns with visual features to achieve a significantly improved classifier. Finally, we propose a learning-to-rank based mutual promotion procedure to incrementally optimize the classifiers based on the unlabeled data in the target domain. Comprehensive experiments based on multiple real surveillance datasets are conducted, and the results show that our algorithm gains significant improvement compared with the state-of-art cross-dataset unsupervised person re-identification algorithms.

cs.CV↗