arXiv subjects

Jian Xu

Publications and source records attributed to Jian Xu.

At least 307 records · Page 17Linked to original sources

arXiv2018

Unsupervised Semantic-based Aggregation of Deep Convolutional Features

In this paper, we propose a simple but effective semantic-based aggregation (SBA) method. The proposed SBA utilizes the discriminative filters of deep convolutional layers as semantic detectors. Moreover, we propose the effective unsupervised strategy to select some semantic detectors to generate the "probabilistic proposals", which highlight certain discriminative pattern of objects and suppress the noise of background. The final global SBA representation could then be acquired by aggregating the regional representations weighted by the selected "probabilistic proposals" corresponding to various semantic content. Our unsupervised SBA is easy to generalize and achieves excellent performance on various tasks. We conduct comprehensive experiments and show that our unsupervised SBA outperforms the state-of-the-art unsupervised and supervised aggregation methods on image retrieval, place recognition and cloud classification.

cs.CV↗

arXiv2018

High-throughput high-resolution multi-scale three dimensional laser printing

Throughput and resolution are both of critical importance for modern manufacturing, however, they are usually contradictory with each other. Here, a high-throughput high-resolution three dimensional (3D) printing is demonstrated by incorporating a simultaneous spatiotemporal focusing (SSTF) scheme in two-photon polymerization (TPP). Remarkably, the SSTF can ensure generation of a spherical focal spot of which the size depends linearly on the laser power. Thus, the resolution can be continuously adjusted from sub-10 μm to ~40 μm by only increasing the laser power. A multi-scale 3D structure is fabricated using this technique, for which the regions of coarse and fine feature sizes are produced at high and low resolutions, respectively.

physics.app-ph↗

arXiv2018

A Bio-inspired Redundant Sensing Architecture

Sensing is the process of deriving signals from the environment that allows artificial systems to interact with the physical world. The Shannon theorem specifies the maximum rate at which information can be acquired. However, this upper bound is hard to achieve in many man-made systems. The biological visual systems, on the other hand, have highly efficient signal representation and processing mechanisms that allow precise sensing. In this work, we argue that redundancy is one of the critical characteristics for such superior performance. We show architectural advantages by utilizing redundant sensing, including correction of mismatch error and significant precision enhancement. For a proof-of-concept demonstration, we have designed a heuristic-based analog-to-digital converter - a zero-dimensional quantizer. Through Monte Carlo simulation with the error probabilistic distribution as a priori, the performance approaching the Shannon limit is feasible. In actual measurements without knowing the error distribution, we observe at least 2-bit extra precision. The results may also help explain biological processes including the dominance of binocular vision, the functional roles of the fixational eye movements, and the structural mechanisms allowing hyperacuity.

cs.NE↗

arXiv2018

Advancing System Performance with Redundancy: From Biological to Artificial Designs

Redundancy is a fundamental characteristic of many biological processes such as those in the genetic, visual, muscular and nervous system; yet its function has not been fully understood. The conventional interpretation of redundancy is that it serves as a fault-tolerance mechanism, which leads to redundancy's de facto application in man-made systems for reliability enhancement. On the contrary, our previous works have demonstrated an example where redundancy can be engineered solely for enhancing other aspects of the system, namely accuracy and precision. This design was inspired by the binocular structure of the human vision which we believe may share a similar operation. In this paper, we present a unified theory describing how such utilization of redundancy is feasible through two complementary mechanisms: representational redundancy (RPR) and entangled redundancy (ETR). Besides the previous works, we point out two additional examples where our new understanding of redundancy can be applied to justify a system's superior performance. One is the human musculoskeletal system (HMS) - a biological instance, and one is the deep residual neural network (ResNet) - an artificial counterpart. We envision that our theory would provide a framework for the future development of bio-inspired redundant artificial systems as well as assist the studies of the fundamental mechanisms governing various biological processes.

cs.IT↗

arXiv2017

Representing Big Data as Networks: New Methods and Insights

Our world produces massive data every day; they exist in diverse forms, from pairwise data and matrix to time series and trajectories. Meanwhile, we have access to the versatile toolkit of network analysis. Networks also have different forms; from simple networks to higher-order network, each representation has different capabilities in carrying information. For researchers who want to leverage the power of the network toolkit, and apply it beyond networks data to sequential data, diffusion data, and many more, the question is: how to represent big data and networks? This dissertation makes a first step to answering the question. It proposes the higher-order network, which is a critical piece for representing higher-order interaction data; it introduces a scalable algorithm for building the network, and visualization tools for interactive exploration. Finally, it presents broad applications of the higher-order network in the real-world.

cs.SI↗

arXiv2017

Long-time asymptotic behavior for the complex short pulse equation

In this paper, we consider the initial value problem for the complex short pulse equation with a Wadati-Konno-Ichikawa type Lax pair. We show that the solution to the initial value problem has a parametric expression in terms of the solution of $2\times 2$-matrix Riemann-Hilbert problem, from which an implicit one-soliton solution is obtained on the discrete spectrum. While on the continuous spectrum we further establish the explicit long-time asymptotic behavior of the non-soliton solution by using Deift-Zhou nonlinear steepest descent method.

math-ph↗

arXiv2017

Centimeter-scale superfine three-dimensional printing with femtosecond laser two-photon polymerization

Nowadays three-dimensional (3D) printing has been widely used for producing geometrically complex 3D structures from a broad range of materials such as ceramics, metals, polymers, semiconductors, etc. Although it has been demonstrated that a fabrication resolution as high as ~100 nm can be achieved in 3D printing based on two photon polymerization (TPP), the end product size of TPP is typically on millimeter scale limited by the short working distance of high-numerical-aperture focal lens. Here we present a method based on simultaneous spatiotemporal focusing (SSTF) of the femtosecond laser pulses that enables to fabricate centimeter-scale 3D structures of fine features with TPP. We also demonstrate an isotropic spatial resolution which can be continuously tuned in the range of ~10 um and ~40 um by only varying the power of femtosecond laser, making this technique extremely flexible and easy to implement. We fabricate several Chinese guardian lions of a maximum height of 0.6 cm and a Terra Cotta Warrior of a height of 1.3 cm using this method.

physics.app-ph↗

arXiv2017

Unsupervised Part-based Weighting Aggregation of Deep Convolutional Features for Image Retrieval

In this paper, we propose a simple but effective semantic part-based weighting aggregation (PWA) for image retrieval. The proposed PWA utilizes the discriminative filters of deep convolutional layers as part detectors. Moreover, we propose the effective unsupervised strategy to select some part detectors to generate the "probabilistic proposals", which highlight certain discriminative parts of objects and suppress the noise of background. The final global PWA representation could then be acquired by aggregating the regional representations weighted by the selected "probabilistic proposals" corresponding to various semantic content. We conduct comprehensive experiments on four standard datasets and show that our unsupervised PWA outperforms the state-of-the-art unsupervised and supervised aggregation methods. Code is available at https://github.com/XJhaoren/PWA.

cs.CV↗

arXiv2017

Structural Diversity and Homophily: A Study Across More than One Hundred Big Networks

A widely recognized organizing principle of networks is structural homophily, which suggests that people with more common neighbors are more likely to connect with each other. However, what influence the diverse structures embedded in common neighbors have on link formation is much less well-understood. To explore this problem, we begin by characterizing the structural diversity of common neighborhoods. Using a collection of 120 large-scale networks, we demonstrate that the impact of the common neighborhood diversity on link existence can vary substantially across networks. We find that its positive effect on Facebook and negative effect on LinkedIn suggest different underlying networking needs in these networks. We also discover striking cases where diversity violates the principle of homophily---that is, where fewer mutual connections may lead to a higher tendency to link with each other. We then leverage structural diversity to develop a common neighborhood signature (CNS), which we apply to a large set of networks to uncover unique network superfamilies not discoverable by conventional methods. Our findings shed light on the pursuit to understand the ways in which network structures are organized and formed, pointing to potential advancement in designing graph generation models and recommender systems.

cs.SI↗

arXiv2017

Initial-boundary value problem for the two-component Gerdjikov-Ivanov equation on the interval

In this paper, we apply Fokas unified method to study initial-boundary value problems for the two-component Gerdjikov-Ivanov equation formulated on the finite interval with $3 \times 3$ Lax pairs. The solution can be expressed in terms of the solution of a $3\times3$ Riemann-Hilbert problem. The relevant jump matrices are explicitly given in terms of three matrix-value spectral functions $s(λ)$, $S(λ)$ and $S_L(λ)$, which arising from the initial values at $t=0$, boundary values at $x=0$ and boundary values at $x=L$, respectively. Moreover, The associated Dirichlet to Neumann map is analyzed via the global relation. The relevant formulae for boundary value problems on the finite interval can reduce to ones on the half-line as the length of the interval tends to infinity.

nlin.SI↗

arXiv2017

Economic Dispatch Considering Spatial and Temporal Correlations of Multiple Renewable Power Plants

The correlations of multiple renewable power plants (RPPs) should be fully considered in the power system with very high penetration renewable power integration. This paper models the uncertainties, spatial correlation of multiple RPPs based on Copula theory and actual probability historical histograms by one-dimension distributions for economic dispatch (ED) problem. An efficient dynamic renewable power scenario generation method based on Gibbs sampling is proposed to generate renewable power scenarios considering the uncertainties, spatial correlation and variability (temporal correlation) of multiple RPPs, in which the sampling space complexity do not increase with the number of RPPs. Distribution-based and scenario-based methods are proposed and compared to solve the real-time ED problem with multiple RPPs. Results show that the proposed dynamic scenario generation method is much more consist with the actual renewable power. The proposed ED methods show better understanding for the uncertainties, spatial and temporal correlations of renewable power and more economical compared with the traditional ones.

math.OC↗

arXiv2017

Stochastic Dynamic Optimal Power Flow in Distribution Network with Distributed Renewable Energy and Battery Energy Storage

The penetration of distributed renewable energy (DRE) greatly raises the risk of distribution network operation such as peak shaving and voltage stability. Battery energy storage (BES) has been widely accepted as the most potential application to cope with the challenge of high penetration of DRE. To cope with the uncertainties and variability of DRE, a stochastic day-ahead dynamic optimal power flow (DOPF) and its algorithm are proposed. The overall economy is achieved by fully considering the DRE, BES, electricity purchasing and active power losses. The rainflow algorithm-based cycle counting method of BES is incorporated in the DOPF model to capture the cell degradation, greatly extending the expected BES lifetime and achieving a better economy. DRE scenarios are generated to consider the uncertainties and correlations based on the Copula theory. To solve the DOPF model, we propose a Lagrange relaxation-based algorithm, which has a significantly reduced complexity with respect to the existing techniques. For this reason, the proposed algorithm enables much more scenarios incorporated in the DOPF model and better captures the DRE uncertainties and correlations. Finally, numerical studies for the day-ahead DOPF in the IEEE 123-node test feeder are presented to demonstrate the merits of the proposed method. Results show that the actual BES life expectancy of the proposed model has increased to 4.89 times compared with the traditional ones. The problems caused by DRE are greatly alleviated by fully capturing the uncertainties and correlations with the proposed method.

math.OC↗

arXiv2017

Mining Features Associated with Effective Tweets

What tweet features are associated with higher effectiveness in tweets? Through the mining of 122 million engagements of 2.5 million original tweets, we present a systematic review of tweet time, entities, composition, and user account features. We show that the relationship between various features and tweeting effectiveness is non-linear; for example, tweets that use a few hashtags have higher effectiveness than using no or too many hashtags. This research closely relates to various industrial applications that are based on tweet features, including the analysis of advertising campaigns, the prediction of user engagement, the extraction of signals for automated trading, etc.

cs.SI↗

arXiv2017

HoNVis: Visualizing and Exploring Higher-Order Networks

Unlike the conventional first-order network (FoN), the higher-order network (HoN) provides a more accurate description of transitions by creating additional nodes to encode higher-order dependencies. However, there exists no visualization and exploration tool for the HoN. For applications such as the development of strategies to control species invasion through global shipping which is known to exhibit higher-order dependencies, the existing FoN visualization tools are limited. In this paper, we present HoNVis, a novel visual analytics framework for exploring higher-order dependencies of the global ocean shipping network. Our framework leverages coordinated multiple views to reveal the network structure at three levels of detail (i.e., the global, local, and individual port levels). Users can quickly identify ports of interest at the global level and specify a port to investigate its higher-order nodes at the individual port level. Investigating a larger-scale impact is enabled through the exploration of HoN at the local level. Using the global ocean shipping network data, we demonstrate the effectiveness of our approach with a real-world use case conducted by domain experts specializing in species invasion. Finally, we discuss the generalizability of this framework to other real-world applications such as information diffusion in social networks and epidemic spreading through air transportation.

cs.GR↗

arXiv2017

Low Power SI Class E Power Amplifier and RF Switch For Health Care

This research was to design a 2.4 GHz class E Power Amplifier (PA) for health care, with 0.18um Semiconductor Manufacturing International Corporation CMOS technology by using Cadence software. And also RF switch was designed at cadence software with power Jazz 180nm SOI process. The ultimate goal for such application is to reach high performance and low cost, and between high performance and low power consumption design. This paper introduces the design of a 2.4GHz class E power amplifier and RF switch design. PA consists of cascade stage with negative capacitance. This power amplifier can transmit 16dBm output power to a 50Ω load. The performance of the power amplifier and switch meet the specification requirements of the desired.

cs.OH↗

arXiv2016

Detecting topological exceptional points in a parity-time symmetric system with cold atoms

We reveal a novel topological property of the exceptional points in a two-level parity-time symmetric system and then propose a scheme to detect the topological exceptional points in the system, which is embedded in a larger Hilbert space constructed by a four-level cold atomic system. We show that a tunable parameter in the presented system for simulating the non-Hermitian Hamiltonian can be tuned to swept the eigenstates through the exceptional points in parameter space. The non-trivial Berry phases of the eigenstates obtained in this loop from the exceptional points can be measured by the atomic interferometry. Since the proposed operations and detection are experimentally feasible, our scheme may pave a promising way to explore the novel properties of non-Hermitian systems.

quant-ph↗

arXiv2016

Long-time asymptotics for the short pulse equation

In this paper, we analyze the long-time behavior of the solution of the initial value problem (IVP) for the short pulse (SP) equation. As the SP equation is a complete integrable system, which posses a Wadati-Konno-Ichikawa (WKI)-type Lax pair, we formulate a $2\times 2$ matrix Riemann-Hilbert problem to this IVP by using the inverse scattering method. Since the spectral variable $k$ is the same order in the WKI-type Lax pair, we construct the solution of this IVP parametrically in the new scale $(y,t)$, whereas the original scale $(x,t)$ is given in terms of functions in the new scale, in terms of the solution of this Riemann-Hilbert problem. However, by employing the nonlinear steepest descent method of Deift and Zhou for oscillatory Riemann-Hilbert problem, we can get the explicit leading order asymptotic of the solution of the short pulse equation in the original scale $(x,t)$ as time $t$ goes to infinity.

nlin.SI↗

arXiv2016

Representing higher-order dependencies in networks

To ensure the correctness of network analysis methods, the network (as the input) has to be a sufficiently accurate representation of the underlying data. However, when representing sequential data from complex systems such as global shipping traffic or web clickstream traffic as networks, conventional network representations that implicitly assume the Markov property (first-order dependency) can quickly become limiting. This assumption holds that when movements are simulated on the network, the next movement depends only on the current node, discounting the fact that the movement may depend on several previous steps. However, we show that data derived from many complex systems can show up to fifth-order dependencies. In these cases, the oversimplifying assumption of the first-order network representation can lead to inaccurate network analysis results. To address this problem, we propose the Higher-Order Network (HON) representation that can discover and embed variable orders of dependencies in a network representation. Through a comprehensive empirical evaluation and analysis, we establish several desirable characteristics of HON, including accuracy, scalability, and direct compatibility with the existing suite of network analysis methods. We illustrate how HON can be applied to a broad variety of tasks, such as random walking, clustering, and ranking, and we demonstrate that by using it as input, HON yields more accurate results without any modification to these tasks.

cs.SI↗