arXiv subjects

Xi Wang

Publications and source records attributed to Xi Wang.

At least 415 records · Page 23Linked to original sources

arXiv2019

Center of circle after perspective transformation

Video-based glint-free eye tracking commonly estimates gaze direction based on the pupil center. The boundary of the pupil is fitted with an ellipse and the euclidean center of the ellipse in the image is taken as the center of the pupil. However, the center of the pupil is generally not mapped to the center of the ellipse by the projective camera transformation. This error resulting from using a point that is not the true center of the pupil directly affects eye tracking accuracy. We investigate the underlying geometric problem of determining the center of a circular object based on its projective image. The main idea is to exploit two concentric circles -- in the application scenario these are the pupil and the iris. We show that it is possible to computed the center and the ratio of the radii from the mapped concentric circles with a direct method that is fast and robust in practice. We evaluate our method on synthetically generated data and find that it improves systematically over using the center of the fitted ellipse. Apart from applications of eye tracking we estimate that our approach will be useful in other tracking applications.

cs.CV↗

arXiv2019

Vortex conveyor belt for matter-wave coherent splitting and interferometry

We numerically study a matter wave interferometer realized by splitting a trapped Bose-Einstein condensate with phase imprinting. We show that a simple step-like imprinting pattern rapidly decays into a string of vortices that can generate opposite velocities on the two halves of the condensate. We first study in detail the splitting and launching effect of these vortex structures, whose functioning resembles the one of a conveyor belt, and we show that the initial exit velocity along the vortex conveyor belt can be controlled continuously by adjusting the vortex distance. We finally characterize the complete interferometric sequence, demonstrating how the phase of the resulting interference fringe can be used to measure an external acceleration. The proposed scheme has the potential to be developed into compact and high precision accelerometers.

cond-mat.quant-gas↗

arXiv2019

QoS-aware Dynamic Fog Service Provisioning

Recent advances in the areas of Internet of Things (IoT), Big Data, and Machine Learning have contributed to the rise of a growing number of complex applications. These applications will be data-intensive, delay-sensitive, and real-time as smart devices prevail more in our daily life. Ensuring Quality of Service (QoS) for delay-sensitive applications is a must, and fog computing is seen as one of the primary enablers for satisfying such tight QoS requirements, as it puts compute, storage, and networking resources closer to the user. In this paper, we first introduce FogPlan, a framework for QoS-aware Dynamic Fog Service Provisioning (QDFSP). QDFSP concerns the dynamic deployment of application services on fog nodes, or the release of application services that have previously been deployed on fog nodes, in order to meet low latency and QoS requirements of applications while minimizing cost. FogPlan framework is practical and operates with no assumptions and minimal information about IoT nodes. Next, we present a possible formulation (as an optimization problem) and two efficient greedy algorithms for addressing the QDFSP at one instance of time. Finally, the FogPlan framework is evaluated using a simulation based on real-world traffic traces.

cs.NI↗

arXiv2018

Deep Learning for micro-Electrocorticographic (μECoG) Data

Machine learning can extract information from neural recordings, e.g., surface EEG, ECoG and μECoG, and therefore plays an important role in many research and clinical applications. Deep learning with artificial neural networks has recently seen increasing attention as a new approach in brain signal decoding. Here, we apply a deep learning approach using convolutional neural networks to μECoG data obtained with a wireless, chronically implanted system in an ovine animal model. Regularized linear discriminant analysis (rLDA), a filter bank component spatial pattern (FBCSP) algorithm and convolutional neural networks (ConvNets) were applied to auditory evoked responses captured by μECoG. We show that compared with rLDA and FBCSP, significantly higher decoding accuracy can be obtained by ConvNets trained in an end-to-end manner, i.e., without any predefined signal features. Deep learning thus proves a promising technique for μECoG-based brain-machine interfacing applications.

eess.SP↗

arXiv2017

Maps of Visual Importance

The importance of an element in a visual stimulus is commonly associated with the fixations during a free-viewing task. We argue that fixations are not always correlated with attention or awareness of visual objects. We suggest to filter the fixations recorded during exploration of the image based on the fixations recorded during recalling the image against a neutral background. This idea exploits that eye movements are a spatial index into the memory of a visual stimulus. We perform an experiment in which we record the eye movements of 30 observers during the presentation and recollection of 100 images. The locations of fixations during recall are only qualitatively related to the fixations during exploration. We develop a deformation mapping technique to align the fixations from recall with the fixation during exploration. This allows filtering the fixations based on proximity and a threshold on proximity provides a convenient slider to control the amount of filtering. Analyzing the spatial histograms resulting from the filtering procedure as well as the set of removed fixations shows that certain types of scene elements, which could be considered irrelevant, are removed. In this sense, they provide a measure of importance of visual elements for human observers.

q-bio.NC↗

arXiv2017

Profitability of simple stationary technical trading rules with high-frequency data of Chinese Index Futures

Technical trading rules have been widely used by practitioners in financial markets for a long time. The profitability remains controversial and few consider the stationarity of technical indicators used in trading rules. We convert MA, KDJ and Bollinger bands into stationary processes and investigate the profitability of these trading rules by using 3 high-frequency data(15s,30s and 60s) of CSI300 Stock Index Futures from January 4th 2012 to December 31st 2016. Several performance and risk measures are adopted to assess the practical value of all trading rules directly while ADF-test is used to verify the stationarity and SPA test to check whether trading rules perform well due to intrinsic superiority or pure luck. The results show that there are several significant combinations of parameters for each indicator when transaction costs are not taken into consideration. Once transaction costs are included, trading profits will be eliminated completely. We also propose a method to reduce the risk of technical trading rules.

q-fin.ST↗

arXiv2017

Large-Scale Domain Adaptation via Teacher-Student Learning

High accuracy speech recognition requires a large amount of transcribed data for supervised training. In the absence of such data, domain adaptation of a well-trained acoustic model can be performed, but even here, high accuracy usually requires significant labeled data from the target domain. In this work, we propose an approach to domain adaptation that does not require transcriptions but instead uses a corpus of unlabeled parallel data, consisting of pairs of samples from the source domain of the well-trained model and the desired target domain. To perform adaptation, we employ teacher/student (T/S) learning, in which the posterior probabilities generated by the source-domain model can be used in lieu of labels to train the target-domain model. We evaluate the proposed approach in two scenarios, adapting a clean acoustic model to noisy speech and adapting an adults speech acoustic model to children speech. Significant improvements in accuracy are obtained, with reductions in word error rate of up to 44% over the original source model without the need for transcribed data in the target domain. Moreover, we show that increasing the amount of unlabeled data results in additional model robustness, which is particularly beneficial when using simulated training data in the target-domain.

cs.CL↗

arXiv2017

Aggregating Frame-level Features for Large-Scale Video Classification

This paper introduces the system we developed for the Google Cloud & YouTube-8M Video Understanding Challenge, which can be considered as a multi-label classification problem defined on top of the large scale YouTube-8M Dataset. We employ a large set of techniques to aggregate the provided frame-level feature representations and generate video-level predictions, including several variants of recurrent neural networks (RNN) and generalized VLAD. We also adopt several fusion strategies to explore the complementarity among the models. In terms of the official metric GAP@20 (global average precision at 20), our best fusion model attains 0.84198 on the public 50\% of test data and 0.84193 on the private 50\% of test data, ranking 4th out of 650 teams worldwide in the competition.

cs.CV↗

arXiv2017

Interaction-driven Quantum Anomalous Hall Effect in Halogenated Hematite Nanosheets

Based on first-principle calculations and $k\cdot p$ model analysis, we show that the quantum anomalous Hall (QAH) insulating phase can be realized in the functionalized hematite (or $α$-Fe$_2$O$_3$) nanosheet and the obtained topological gap can be as large as $\sim$300 meV. The driving force of the topological phase is the strong interactions of localized Fe 3$d$ electrons operating on the quadratic band crossing point of the non-interacting band structures. Such interaction driven QAH insulator is different from the single particle band topology mechanism in experimentally realized QAH insulator, the magnetic ion doped topological insulator film. Depending on the thickness of the nanosheet, topological insulating state with helical-like or chiral edge states can be realized. Our work provides a realization of the interaction-driven QAH insulating state in a realistic material.

cond-mat.mes-hall↗

arXiv2016

Self-Assembled, Nanostructured, Tunable Metamaterials via Spinodal Decomposition

Self-assembly via nanoscale phase-separation offers an elegant route to fabricate nanocomposites with physical properties unattainable in single-component systems. One important class of nanocomposites are optical metamaterials which exhibit exotic properties and lead to opportunities for agile control of light propagation. Such metamaterials are typically fabricated via expensive and hard-to-scale top-down processes requiring precise integration of dissimilar materials. In turn, there is a need for alternative, more efficient routes to fabricate large-scale metamaterials for practical applications with deep-subwavelength resolution. Here, we demonstrate a bottom-up approach to fabricate scalable nanostructured metamaterials via spinodal decomposition. To demonstrate the potential of such an approach, we leverage the innate spinodal decomposition of the VO2-TiO2 system, the metal-to-insulator transition in VO2, and thin-film epitaxy, to produce self-organized nanostructures with coherent interfaces and a structural unit cell down to 15 nm (tunable between horizontally- and vertically-aligned lamellae) wherein the iso-frequency surface is temperature-tunable from elliptic- to hyperbolic-dispersion producing metamaterial behavior. These results provide an efficient route for the fabrication of nanostructured metamaterials and other nanocomposites for desired functionalities.

cond-mat.mtrl-sci↗

arXiv2015

Fusing Multi-Stream Deep Networks for Video Classification

This paper studies deep network architectures to address the problem of video classification. A multi-stream framework is proposed to fully utilize the rich multimodal information in videos. Specifically, we first train three Convolutional Neural Networks to model spatial, short-term motion and audio clues respectively. Long Short Term Memory networks are then adopted to explore long-term temporal dynamics. With the outputs of the individual streams, we propose a simple and effective fusion method to generate the final predictions, where the optimal fusion weights are learned adaptively for each class, and the learning process is regularized by automatically estimated class relationships. Our contributions are two-fold. First, the proposed multi-stream framework is able to exploit multimodal features that are more comprehensive than those previously attempted. Second, we demonstrate that the adaptive fusion method using the class relationship as a regularizer outperforms traditional alternatives that estimate the weights in a "free" fashion. Our framework produces significantly better results than the state of the arts on two popular benchmarks, 92.2\% on UCF-101 (without using audio) and 84.9\% on Columbia Consumer Videos.

cs.CV↗

arXiv2015

Hamiltonian Properties of DCell Networks

DCell has been proposed for data centers as a server centric interconnection network structure. DCell can support millions of servers with high network capacity by only using commodity switches. With one exception, we prove that a $k$ level DCell built with $n$ port switches is Hamiltonian-connected for $k \geq 0$ and $n \geq 2$. Our proof extends to all generalized DCell connection rules for $n\ge 3$. Then, we propose an $O(t_k)$ algorithm for finding a Hamiltonian path in $DCell_{k}$, where $t_k$ is the number of servers in $DCell_{k}$. What's more, we prove that $DCell_{k}$ is $(n+k-4)$-fault Hamiltonian-connected and $(n+k-3)$-fault Hamiltonian. In addition, we show that a partial DCell is Hamiltonian connected if it conforms to a few practical restrictions.

cs.DC↗

arXiv2015

Evaluating Two-Stream CNN for Video Classification

Videos contain very rich semantic information. Traditional hand-crafted features are known to be inadequate in analyzing complex video semantics. Inspired by the huge success of the deep learning methods in analyzing image, audio and text data, significant efforts are recently being devoted to the design of deep nets for video analytics. Among the many practical needs, classifying videos (or video clips) based on their major semantic categories (e.g., "skiing") is useful in many applications. In this paper, we conduct an in-depth study to investigate important implementation options that may affect the performance of deep nets on video classification. Our evaluations are conducted on top of a recent two-stream convolutional neural network (CNN) pipeline, which uses both static frames and motion optical flows, and has demonstrated competitive performance against the state-of-the-art methods. In order to gain insights and to arrive at a practical guideline, many important options are studied, including network architectures, model fusion, learning parameters and the final prediction methods. Based on the evaluations, very competitive results are attained on two popular video classification benchmarks. We hope that the discussions and conclusions from this work can help researchers in related fields to quickly set up a good basis for further investigations along this very promising direction.

cs.CV↗

arXiv2015

Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification

Classifying videos according to content semantics is an important problem with a wide range of applications. In this paper, we propose a hybrid deep learning framework for video classification, which is able to model static spatial information, short-term motion, as well as long-term temporal clues in the videos. Specifically, the spatial and the short-term motion features are extracted separately by two Convolutional Neural Networks (CNN). These two types of CNN-based features are then combined in a regularized feature fusion network for classification, which is able to learn and utilize feature relationships for improved performance. In addition, Long Short Term Memory (LSTM) networks are applied on top of the two features to further model longer-term temporal clues. The main contribution of this work is the hybrid learning framework that can model several important aspects of the video data. We also show that (1) combining the spatial and the short-term motion features in the regularized fusion network is better than direct classification and fusion using the CNN with a softmax layer, and (2) the sequence-based LSTM is highly complementary to the traditional classification strategy without considering the temporal frame orders. Extensive experiments are conducted on two popular and challenging benchmarks, the UCF-101 Human Actions and the Columbia Consumer Videos (CCV). On both benchmarks, our framework achieves to-date the best reported performance: $91.3\%$ on the UCF-101 and $83.5\%$ on the CCV.

cs.CV↗

arXiv2015

Large-angle quasi-self-collimation effect in a rod-type photonic crystal

A rod-type photonic crystal (PC) with a rectangular lattice shows a large-angle quasi-self-collimation (quasi-SC) effect by changing the symmetry of its rectangular lattice to straighten one of the isofrequency contours. To investigate the straightness of the isofrequency contour as well as the quasi-SC effect, we propose a straightness factor L based on the method of least squares. With L smaller than L0 (L0 = 0.01 is the critical value), the isofrequency contour is sufficiently straight to induce quasi-SC effect with the beam quasi-collimating in the structure. Furthermore, the efficiency of light coupling to the quasi-SC PC is studied, and can be greatly improved by applying a carefully designed antireflection layer.

physics.optics↗

arXiv2014

Rapidly reconfigurable radio-frequency arbitrary waveforms synthesized on a CMOS photonic chip

Photonic methods of radio-frequency waveform generation and processing provide performance and flexibility over electronic methods due to the ultrawide bandwidth offered by the optical carriers. However, they suffer from lack of integration and slow reconfiguration speed. Here we propose an architecture of integrated photonic RF waveform generation and processing, and implement it on a silicon chip fabricated in a semiconductor manufacturing foundry. Our device can generate programmable RF bursts or continuous waveforms with only the light source, electrical drives/controls and detectors being off chip. It turns on and off an individual pulse in the RF burst within 4 nanoseconds, achieving a reconfiguration speed three orders of magnitude faster than thermal tuning. The on-chip optical delay elements offers an integrated approach to accurately manipulate individual RF waveform features without constrains set by the speed and timing jitter of electronics, and should find broad applications ranging from high-speed wireless to defense electronics.

physics.optics↗

arXiv2014

User recommendation in reciprocal and bipartite social networks -- a case study of online dating

Many social networks in our daily life are bipartite networks built on reciprocity. How can we recommend users/friends to a user, so that the user is interested in and attractive to recommended users? In this research, we propose a new collaborative filtering model to improve user recommendations in reciprocal and bipartite social networks. The model considers a user's "taste" in picking others and "attractiveness" in being picked by others. A case study of an online dating network shows that the new model has good performance in recommending both initial and reciprocal contacts.

cs.SI↗

arXiv2013

Millimeter-scale and large-angle self-collimation in a photonic crystal composed of silicon nanorods

We report the observation of a large-angle self-collimation phenomenon occurring in photonic crystals composed of nanorods. Electromagnetic waves incident onto such photonic crystals from directions covering a wide-range of incident angles become highly localized along a single array of rods, which results in narrow-beam propagation without divergence. A propagation length of 0.4 mm is experimentally observed over the wavelength range of 1540 nm to 1570 nm, even in the large incident angle case, which is a very considerable length scale for on-chip optical interconnection.

physics.optics↗