Searcharxiv⌕ Search

arXiv subjects

Xian Wu

Publications and source records attributed to Xian Wu.

208 records · Page 12Linked to original sources
arXiv2017

Content-Adaptive Sketch Portrait Generation by Decompositional Representation Learning

Sketch portrait generation benefits a wide range of applications such as digital entertainment and law enforcement. Although plenty of efforts have been dedicated to this task, several issues still remain unsolved for generating vivid and detail-preserving personal sketch portraits. For example, quite a few artifacts may exist in synthesizing hairpins and glasses, and textural details may be lost in the regions of hair or mustache. Moreover, the generalization ability of current systems is somewhat limited since they usually require elaborately collecting a dictionary of examples or carefully tuning features/components. In this paper, we present a novel representation learning framework that generates an end-to-end photo-sketch mapping through structure and texture decomposition. In the training stage, we first decompose the input face photo into different components according to their representational contents (i.e., structural and textural parts) by using a pre-trained Convolutional Neural Network (CNN). Then, we utilize a Branched Fully Convolutional Neural Network (BFCN) for learning structural and textural representations, respectively. In addition, we design a Sorted Matching Mean Square Error (SM-MSE) metric to measure texture patterns in the loss function. In the stage of sketch rendering, our approach automatically generates structural and textural representations for the input photo and produces the final result via a probabilistic fusion scheme. Extensive experiments on several challenging benchmarks suggest that our approach outperforms example-based synthesis algorithms in terms of both perceptual and objective metrics. In addition, the proposed method also has better generalization ability across dataset without additional training.

cs.CV↗
arXiv2015

Characterization of a gate-defined double quantum dot in a Si/SiGe nanomembrane

We report the fabrication and characterization of a gate-defined double quantum dot formed in a Si/SiGe nanomembrane. In the past, all gate-defined quantum dots in Si/SiGe heterostructures were formed on top of strain-graded virtual substrates. The strain grading process necessarily introduces misfit dislocations into a heterostructure, and these defects introduce lateral strain inhomogeneities, mosaic tilt, and threading dislocations. The use of a SiGe nanomembrane as the virtual substrate enables the strain relaxation to be entirely elastic, eliminating the need for misfit dislocations. However, in this approach the formation of the heterostructure is more complicated, involving two separate epitaxial growth procedures separated by a wet-transfer process that results in a buried non-epitaxial interface 625 nm from the quantum dot. We demonstrate that in spite of this buried interface in close proximity to the device, a double quantum dot can be formed that is controllable enough to enable tuning of the inter-dot tunnel coupling, the identification of spin states, and the measurement of a singlet-to-triplet transition as a function of an applied magnetic field.

cond-mat.mes-hall↗
arXiv2015

Low-noise kinetic inductance traveling-wave amplifier using three-wave mixing

We have fabricated a wide-bandwidth, high dynamic range, low-noise cryogenic amplifier based on a superconducting kinetic inductance traveling-wave device. The device was made from NbTiN and consisted of a long, coplanar waveguide on a silicon chip. By adding a DC current and an RF pump tone we are able to generate parametric amplification using three-wave mixing. The devices exhibit gain of more than 15 dB across an instantaneous bandwidth from 4 to 8 GHz. The total usable gain bandwidth, including both sides of the signal-idler gain region, is more than 6 GHz. The noise referred to the input of the devices approaches the quantum limit, with less than 1 photon excess noise. Compared to similarly constructed four-wave mixing amplifiers, these devices operate with the RF pump at $\sim$20 dB lower power and at frequencies far from the signal. This will permit easier integration into large scale qubit and detector applications.

cond-mat.supr-con↗
arXiv2015

End-to-End Photo-Sketch Generation via Fully Convolutional Representation Learning

Sketch-based face recognition is an interesting task in vision and multimedia research, yet it is quite challenging due to the great difference between face photos and sketches. In this paper, we propose a novel approach for photo-sketch generation, aiming to automatically transform face photos into detail-preserving personal sketches. Unlike the traditional models synthesizing sketches based on a dictionary of exemplars, we develop a fully convolutional network to learn the end-to-end photo-sketch mapping. Our approach takes whole face photos as inputs and directly generates the corresponding sketch images with efficient inference and learning, in which the architecture are stacked by only convolutional kernels of very small sizes. To well capture the person identity during the photo-sketch transformation, we define our optimization objective in the form of joint generative-discriminative minimization. In particular, a discriminative regularization term is incorporated into the photo-sketch generation, enhancing the discriminability of the generated person sketches against other individuals. Extensive experiments on several standard benchmarks suggest that our approach outperforms other state-of-the-art methods in both photo-sketch generation and face sketch verification.

cs.CV↗
arXiv2014

Two-axis control of a singlet-triplet qubit with an integrated micromagnet

The qubit is the fundamental building block of a quantum computer. We fabricate a qubit in a silicon double quantum dot with an integrated micromagnet in which the qubit basis states are the singlet state and the spin-zero triplet state of two electrons. Because of the micro magnet, the magnetic field difference $ΔB$ between the two sides of the double dot is large enough to enable the achievement of coherent rotation of the qubit's Bloch vector about two different axes of the Bloch sphere. By measuring the decay of the quantum oscillations, the inhomogeneous spin coherence time $T_{2}^{*}$ is determined. By measuring $T_{2}^{*}$ at many different values of the exchange coupling $J$ and at two different values of $ΔB$, we provide evidence that the micromagnet does not limit decoherence, with the dominant limits on $T_{2}^{*}$ arising from charge noise and from coupling to nuclear spins.

cond-mat.mes-hall↗
arXiv2014

The Multi-shop Ski Rental Problem

We consider the {\em multi-shop ski rental} problem. This problem generalizes the classic ski rental problem to a multi-shop setting, in which each shop has different prices for renting and purchasing a pair of skis, and a \emph{consumer} has to make decisions on when and where to buy. We are interested in the {\em optimal online (competitive-ratio minimizing) mixed strategy} from the consumer's perspective. For our problem in its basic form, we obtain exciting closed-form solutions and a linear time algorithm for computing them. We further demonstrate the generality of our approach by investigating three extensions of our basic problem, namely ones that consider costs incurred by entering a shop or switching to another shop. Our solutions to these problems suggest that the consumer must assign positive probability in \emph{exactly one} shop at any buying time. Our results apply to many real-world applications, ranging from cost management in \texttt{IaaS} cloud to scheduling in distributed computing.

cs.GT↗
arXiv2013

Fast coherent manipulation of three-electron states in a double quantum dot

A fundamental goal in the manipulation of quantum systems is the achievement of many coherent oscillations within the characteristic dephasing time T2*[1]. Most manipulations of electron spins in quantum dots have focused on the construction and control of two-state quantum systems, or qubits, in which each quantum dot is occupied by a single electron[2-7]. Here we perform quantum manipulations on a system with more electrons per quantum dot, in a double dot with three electrons. We demonstrate that tailored pulse sequences can be used to induce coherent rotations between 3-electron quantum states. Certain pulse sequences yield coherent oscillations with a very high figure of merit (the ratio of coherence time to rotation time) of >100. The presence of the third electron enables very fast rotations to all possible states, in contrast to the case when only two electrons are used, in which some rotations are slow. The minimum oscillation frequency we observe is >5 GHz.

cond-mat.mes-hall↗
arXiv1994

Residual Intersections and Some Applications

We give a new residual intersection decomposition for the refined intersection products of Fulton-MacPherson. Our formula refines the celebrated residual intersection formula of Fulton, Kleiman, Laksov, and MacPherson. The new decomposition is more likely to be compatible with the canonical decomposition of the intersection products and each term in the decomposition thus has simple geometric meaning. Our study is motivated by its applications to some geometric problems. In particular, we use the decomposition to find the distribution of limiting linear subspaces in degenerations of hypersurfaces. A family of identities for characteristic classes of vector bundles is also obtained as another consequence. This paper will appear in Duke Math. Jour.

alg-geom↗
arXiv1993

Refined intersection products and limiting linear subspaces of hypersurfaces

Let $X$ be a hypersurface of degree $d$ in $\Bbb P^n$ and $F_X$ be the scheme of $\Bbb P^r$'s contained in $X$. If $X$ is generic, then $F_X$ will have the expected dimension (or empty) and its class in the Chow ring of $G(r+1,n+1)$ is given by the top Chern class of the vector bundle $S^dU^*$, where $U$ is the universal subbundle on the Grassmannian $G(r+1,n+1)$. When we deform a generic $X$ into a degenerate $X_0$, the dimension of $F_X$ can jump. In this case, there is a subscheme $F_{lim}$ of $F_{X_0}$ with the expected dimension which consists of limiting $\Bbb P^r$'s in $X_0$ with respect to a general deformation. The simplest example is the well-known case of $27$ lines in a generic cubic surface. If we degenerate the surface into the union of a plane and a quadric, then there are infinitely many lines in the union. Which $27$ lines are the limiting ones and how many of them are in the plane and how many of them are in the quadric? The goal of this paper is to study $F_{lim}$ in general.

alg-geom↗