arXiv subjects

Yi Xu

Publications and source records attributed to Yi Xu.

At least 415 records · Page 23Linked to original sources

arXiv2018

Analogy Search Engine: Finding Analogies in Cross-Domain Research Papers

In recent years, with the rapid proliferation of research publications in the field of Artificial Intelligence, it is becoming increasingly difficult for researchers to effectively keep up with all the latest research in one's own domains. However, history has shown that scientific breakthroughs often come from collaborations of researchers from different domains. Traditional search algorithms like Lexical search, which look for literal matches or synonyms and variants of the query words, are not effective for discovering cross-domain research papers and meeting the needs of researchers in this age of information overflow. In this paper, we developed and tested an innovative semantic search engine, Analogy Search Engine (ASE), for 2000 AI research paper abstracts across domains like Language Technologies, Robotics, Machine Learning, Computational Biology, Human Computer Interactions, etc. ASE combines recent theories and methods from Computational Analogy and Natural Language Processing to go beyond keyword-based lexical search and discover the deeper analogical relationships among research paper abstracts. We experimentally show that ASE is capable of finding more interesting and useful research papers than baseline elasticsearch. Furthermore, we believe that the methods used in ASE go beyond academic paper and will benefit many other document search tasks.

cs.IR↗

arXiv2018

Isotropic Magnetic Purcell Effect

Manipulating the spontaneous emission rate of optical emitters with all-dielectric nanoparticles benefits from their low-loss nature and thus provides relatively large extrinsic quantum yield. However, such Purcell effect greatly depends on the orientation of the dipole emitter. Here, we introduce the concept of isotropic magnetic Purcell effect with Purcell factors about 300 and large extrinsic quantum yield (more than 80%) for a magnetic dipole emitter of arbitrary orientation in an asymmetric silicon nanocavity. The extrinsic quantum yield can be even boosted up to nearly 100% by utilizing a GaP nanocavity. Isotropy of the Purcell factor is manifested via the orientation-independent emission of the magnetic dipole source. This isotropic Purcell effect is robust against small displacement of emitter on the order of 10 nm, releasing the requirement of precise alignment in experiments.

physics.optics↗

arXiv2018

Unidirectional emission in an all-dielectric nanoantenna

All-dielectric nanoantennas are a promising alternative to plasmonic optical antennas for engineering light emission because of their low-loss nature in the optical spectrum. Nevertheless, it is still challenging to manipulate directional light emission with subwavelength all-dielectric nanoantennas. Here, we propose and numerically demonstrate that a hollow silicon nanodisk can serve as a versatile antenna for directing and enhancing the emission from either an electric or magnetic dipole emitter. When primarily coupled to both electric and magnetic dipole modes of a nanoantenna, broadband nearly-unidirectional emission can be realized by the interference of two modes, which can be spectrally tuned via the geometric parameters in an easy way. More importantly, the emission directions for the magnetic and electric dipole emitters are shown as opposite to each other through control of the phase difference between the induced magnetic and electric dipole modes of the antenna. Meanwhile, the Purcell factors can be enhanced by more than one order of magnitude and high quantum efficiencies can be maintained at the visible spectrum for both kinds of dipole emitters. We further show that these unidirectional emission phenomena can withstand small disorder effects of in-plane dipole orientation and location. Our study provides a simple yet versatile platform that can shape the emission of both magnetic and electric dipole emitters.

physics.optics↗

arXiv2018

Frank-Wolfe Method is Automatically Adaptive to Error Bound Condition

Error bound condition has recently gained revived interest in optimization. It has been leveraged to derive faster convergence for many popular algorithms, including subgradient methods, proximal gradient method and accelerated proximal gradient method. However, it is still unclear whether the Frank-Wolfe (FW) method can enjoy faster convergence under error bound condition. In this short note, we give an affirmative answer to this question. We show that the FW method (with a line search for the step size) for optimization over a strongly convex set is automatically adaptive to the error bound condition of the problem. In particular, the iteration complexity of FW can be characterized by $O(\max(1/ε^{1-θ}, \log(1/ε)))$ where $θ\in[0,1]$ is a constant that characterizes the error bound condition. Our results imply that if the constrained set is characterized by a strongly convex function and the objective function can achieve a smaller value outside the considered domain, then the FW method enjoys a fast rate of $O(1/t^2)$.

math.OC↗

arXiv2018

A New Class of Positive Semi-definite Tensors

In this paper, a new class of positive semi-definite tensors, the MO tensor, is introduced. It is inspired by the structure of Moler matrix, a class of test matrices. Then we focus on two special cases in the MO-tensors: Sup-MO tensor and essential MO tensor. They are proved to be positive definite tensors. Especially, the smallest H-eigenvalue of a Sup-MO tensor is positive and tends to zero as the dimension tends to infinity, and an essential MO tensor is also a completely positive tensor.

math.SP↗

arXiv2018

A Weighted Superposition of Functional Contours Model for Modelling Contextual Prominence of Elementary Prosodic Contours

The way speech prosody encodes linguistic, paralinguistic and non-linguistic information via multiparametric representations of the speech signals is still an open issue. The Superposition of Functional Contours (SFC) model proposes to decompose prosody into elementary multiparametric functional contours through the iterative training of neural network contour generators using analysis-by-synthesis. Each generator is responsible for computing multiparametric contours that encode one given linguistic, paralinguistic and non-linguistic information on a variable scope of rhythmic units. The contributions of all generators' outputs are then overlapped and added to produce the prosody of the utterance. We propose an extension of the contour generators that allows them to model the prominence of the elementary contours based on contextual information. WSFC jointly learns the patterns of the elementary multiparametric functional contours and their weights dependent on the contours' contexts. The experimental results show that the proposed weighted SFC (WSFC) model can successfully capture contour prominence and thus improve SFC modelling performance. The WSFC is also shown to be effective at modelling the impact of attitudes on the prominence of functional contours cuing syntactic relations in French, and that of emphasis on the prominence of tone contours in Chinese.

eess.AS↗

arXiv2018

Learning with Non-Convex Truncated Losses by SGD

Learning with a {\it convex loss} function has been a dominating paradigm for many years. It remains an interesting question how non-convex loss functions help improve the generalization of learning with broad applicability. In this paper, we study a family of objective functions formed by truncating traditional loss functions, which is applicable to both shallow learning and deep learning. Truncating loss functions has potential to be less vulnerable and more robust to large noise in observations that could be adversarial. More importantly, it is a generic technique without assuming the knowledge of noise distribution. To justify non-convex learning with truncated losses, we establish excess risk bounds of empirical risk minimization based on truncated losses for heavy-tailed output, and statistical error of an approximate stationary point found by stochastic gradient descent (SGD) method. Our experiments for shallow and deep learning for regression with outliers, corrupted data and heavy-tailed noise further justify the proposed method.

cs.LG↗

arXiv2018

First-order Stochastic Algorithms for Escaping From Saddle Points in Almost Linear Time

Two classes of methods have been proposed for escaping from saddle points with one using the second-order information carried by the Hessian and the other adding the noise into the first-order information. The existing analysis for algorithms using noise in the first-order information is quite involved and hides the essence of added noise, which hinder further improvements of these algorithms. In this paper, we present a novel perspective of noise-adding technique, i.e., adding the noise into the first-order information can help extract the negative curvature from the Hessian matrix, and provide a formal reasoning of this perspective by analyzing a simple first-order procedure. More importantly, the proposed procedure enables one to design purely first-order stochastic algorithms for escaping from non-degenerate saddle points with a much better time complexity (almost linear time in terms of the problem's dimensionality). In particular, we develop a {\bf first-order stochastic algorithm} based on our new technique and an existing algorithm that only converges to a first-order stationary point to enjoy a time complexity of {$\widetilde O(d/ε^{3.5})$ for finding a nearly second-order stationary point $\bf{x}$ such that $\|\nabla F(bf{x})\|\leq ε$ and $\nabla^2 F(bf{x})\geq -\sqrtεI$ (in high probability), where $F(\cdot)$ denotes the objective function and $d$ is the dimensionality of the problem. To the best of our knowledge, this is the best theoretical result of first-order algorithms for stochastic non-convex optimization, which is even competitive with if not better than existing stochastic algorithms hinging on the second-order information.

math.OC↗

arXiv2018

NEON+: Accelerated Gradient Methods for Extracting Negative Curvature for Non-Convex Optimization

Accelerated gradient (AG) methods are breakthroughs in convex optimization, improving the convergence rate of the gradient descent method for optimization with smooth functions. However, the analysis of AG methods for non-convex optimization is still limited. It remains an open question whether AG methods from convex optimization can accelerate the convergence of the gradient descent method for finding local minimum of non-convex optimization problems. This paper provides an affirmative answer to this question. In particular, we analyze two renowned variants of AG methods (namely Polyak's Heavy Ball method and Nesterov's Accelerated Gradient method) for extracting the negative curvature from random noise, which is central to escaping from saddle points. By leveraging the proposed AG methods for extracting the negative curvature, we present a new AG algorithm with double loops for non-convex optimization~\footnote{this is in contrast to a single-loop AG algorithm proposed in a recent manuscript~\citep{AGNON}, which directly analyzed the Nesterov's AG method for non-convex optimization and appeared online on November 29, 2017. However, we emphasize that our work is an independent work, which is inspired by our earlier work~\citep{NEON17} and is based on a different novel analysis.}, which converges to second-order stationary point $\x$ such that $\|\nabla f(\x)\|\leq ε$ and $\nabla^2 f(\x)\geq -\sqrtε I$ with $\widetilde O(1/ε^{1.75})$ iteration complexity, improving that of gradient descent method by a factor of $ε^{-0.25}$ and matching the best iteration complexity of second-order Hessian-free methods for non-convex optimization.

math.OC↗

arXiv2018

Flexible Network Binarization with Layer-wise Priority

How to effectively approximate real-valued parameters with binary codes plays a central role in neural network binarization. In this work, we reveal an important fact that binarizing different layers has a widely-varied effect on the compression ratio of network and the loss of performance. Based on this fact, we propose a novel and flexible neural network binarization method by introducing the concept of layer-wise priority which binarizes parameters in inverse order of their layer depth. In each training step, our method selects a specific network layer, minimizes the discrepancy between the original real-valued weights and its binary approximations, and fine-tunes the whole network accordingly. During the iteration of the above process, it is significant that we can flexibly decide whether to binarize the remaining floating layers or not and explore a trade-off between the loss of performance and the compression ratio of model. The resulting binary network is applied for efficient pedestrian detection. Extensive experimental results on several benchmarks show that under the same compression ratio, our method achieves much lower miss rate and faster detection speed than the state-of-the-art neural network binarization method.

cs.CV↗

arXiv2017

Spectral conditions of complement for some graphical properties

L.H. Feng at el \cite{feng4} present sufficient conditions based on spectral radius for a graph with large minimum degree to be $s$-path-coverable and $s$-Hamiltonian. Motivated by this study, in this paper, we give the sufficient conditions for a graph with large minimum degree to be $s$-connected, $s$-edge-connected, $β$-deficient, $s$-path-coverable, $s$-Hamiltonian and $s$-edge-Hamiltonian in terms of spectral radius of its complement.

math.CO↗

arXiv2017

Design of an optimized $\text{MoS}_2$-based highly sensitive near-infrared surface plasmon resonance imaging biosensor

A surface plasmon resonance imaging biosensor based on $\text{MoS}_2$ deposited on Aluminium substrate is designed for high imaging sensitivity and detection accuracy. The proposed biosensor exhibits better performance than graphene-based biosensor in the near-infrared regime. A high imaging sensitivity of more than 970 $\text{RIU}^{-1}$ is obtained at the wavelength of 1540 nm. The effect of aluminium thickness, number of $\text{MoS}_2$ layers and the refractive index of sensing layer are investigated to obtain an optimized design for high sensor performance. In addition, the sensor performance comparison of $\text{MoS}_2$ and other two-dimensional transition metal dichalcogenide materials based biosensor in the near-infrared regime are also presented. The designed $\text{MoS}_2$ mediated surface plasmon resonance imaging biosensor could provide potential applications in surface plasmon resonance imaging detection of multiple biomolecular interactions simultaneously.

cond-mat.mes-hall↗

arXiv2017

Ultrasensitive and highly accurate long-range surface plasmon resonance biosensors based on two-dimensional transition metal dichalcogenides

Two-dimensional transition metal dichalcogenides (TMDCs), as promising alternative plasmonics supporting materials to graphene, exhibit potential applications in sensing. Here, we propose an ultrasensitive, accurate long-range surface plasmon resonance (LRSPR) imaging biosensor with two-dimensional TMDC layers, which shows higher detection accuracy than that of conventional SPR biosensor. It is found that the imaging sensitivity of the proposed LRSPR biosensor can be enhanced by the integration of TMDC layers, which is different from the previous graphene-based LRSPR or SPR imaging sensor, whose imaging sensitivity usually decreases with the number of graphene layers. The sensitivity enhancement or degradation effect for the proposed chalcogenide-cytop-gold-TMDCs based biosensor depends on the thickness of gold thin film and cytop layer. Imaging sensitivity of more than 4000 $\text{RIU}^{-1}$ can be obtained with a high detection accuracy of more than 120 $\text{deg}^{-1}$. We expect that the proposed TMDCs mediated LRSPR imaging sensor could provide potential applications in chemical sensing and biosensing for a highly sensitive and accurate simultaneous detection of multiple biomolecular interactions.

physics.app-ph↗

arXiv2017

3D Nanoporous Gold with Very Low Parting Limit Derived from Au-based Metallic Glass and Enhanced Methanol Electro-oxidation Catalytic Performance

Nanoporous gold (NPG) with bi-continuous ligaments and pores structure has promising potential in functional applications, among which one prominent example is fuel cell electrocatalyst. However, current application of NPG is mostly limited to methanol electro-oxidation due to its weak catalytic performance. Here we report a simple chemical dealloying process, for generating peculiar three-dimensional (3D) free-standing NPG with high specific surface area associated with a novel porous cone shaped protrusion morphology. This NPG structure possesses the highest specific activity of catalytic performance reported NPG catalysts so far.

cond-mat.mtrl-sci↗

arXiv2017

Enhanced betatron radiation by steering a low-energy-spread electron beam in a deflected laser-driven plasma wiggler

Laser wakefield accelerators (LWFA) hold great potential to produce high-quality high-energy electron beams (e beams) and simultaneously bright x-ray sources via betatron radiation, which are very promising for pump-probe study in ultrafast science. However, in order to obtain a high-quality e beam, electron injection and acceleration should be carefully manipulated, where a large oscillation amplitude has to be avoided and thus the emitted x-ray yield is limited. Here, we report a new scheme to experimentally enhance betatron radiation significantly both in photon yield and photon energy by separating electron injection and acceleration from manipulation of the e-beam transverse oscillation in the wake via introducing a slanted thin plasma refraction slab. Particle-in-cell simulations indicate that the e-beam transverse oscillation amplitude can be increased by more than 10 folds, after being steered into the deflected laser-driven wakefield due to refraction at the slab's boundaries. Spectral broadening of the x-rays can be suppressed owing to the small variation in the peak energy of the low-energy-spread e beam in a plasma wiggler regime. We demonstrate that the high-quality e-beam generation, refracting and wiggling can act as a whole to realize the concurrence of monoenergetic e beam and bright x-rays in a compact LWFA.

physics.plasm-ph↗

arXiv2017

White light emission from silicon nanoparticles

As one of the most important semiconductors, silicon (Si) has been used to fabricate electronic devices, waveguides, detectors, and solar cells etc. However, its indirect bandgap hinders the use of Si for making good emitters1. For integrated photonic circuits, Si-based emitters with sizes in the range of 100-300 nm are highly desirable. Here, we show that efficient white light emission can be realized in spherical and cylindrical Si nanoparticles with feature sizes of ~200 nm. The up-converted luminescence appears at the magnetic and electric multipole resonances when the nanoparticles are resonantly excited at their magnetic and electric dipole resonances by using femtosecond (fs) laser pulses with ultralow low energy of ~40 pJ. The lifetime of the white light is as short as ~52 ps, almost three orders of magnitude smaller than the state-of-the-art results reported so far for Si (~10 ns). Our finding paves the way for realizing efficient Si-based emitters compatible with current semiconductor fabrication technology, which can be integrated to photonic circuits.

physics.optics↗

arXiv2017

Ideal magnetic dipole scattering

We introduce the concept of tunable ideal magnetic dipole scattering, where a nonmagnetic nanoparticle scatters lights as a pure magnetic dipole. High refractive index subwavelength nanoparticles usually support both electric and magnetic dipole responses. Thus, to achieve ideal magnetic dipole scattering one has to suppress the electric dipole response. Such a possibility was recently demonstrated for the so-called anapole mode, which is associated with zero electric dipole scattering. By overlapping magnetic dipole resonance with the anapole mode we achieve ideal magnetic dipole scattering in the far-field with tunable high scattering resonances in near infrared spectrum. We demonstrate that such condition can be realized for two subwavelength geometries. One of them is core-shell nanosphere consisting of Au core and silicon shell. It can be also achieved in other geometries, including nanodisks, which are compatible with current nanofabrication technology.

physics.optics↗

arXiv2016

Efficient Non-oblivious Randomized Reduction for Risk Minimization with Improved Excess Risk Guarantee

In this paper, we address learning problems for high dimensional data. Previously, oblivious random projection based approaches that project high dimensional features onto a random subspace have been used in practice for tackling high-dimensionality challenge in machine learning. Recently, various non-oblivious randomized reduction methods have been developed and deployed for solving many numerical problems such as matrix product approximation, low-rank matrix approximation, etc. However, they are less explored for the machine learning tasks, e.g., classification. More seriously, the theoretical analysis of excess risk bounds for risk minimization, an important measure of generalization performance, has not been established for non-oblivious randomized reduction methods. It therefore remains an open problem what is the benefit of using them over previous oblivious random projection based approaches. To tackle these challenges, we propose an algorithmic framework for employing non-oblivious randomized reduction method for general empirical risk minimizing in machine learning tasks, where the original high-dimensional features are projected onto a random subspace that is derived from the data with a small matrix approximation error. We then derive the first excess risk bound for the proposed non-oblivious randomized reduction approach without requiring strong assumptions on the training data. The established excess risk bound exhibits that the proposed approach provides much better generalization performance and it also sheds more insights about different randomized reduction approaches. Finally, we conduct extensive experiments on both synthetic and real-world benchmark datasets, whose dimension scales to $O(10^7)$, to demonstrate the efficacy of our proposed approach.

cs.LG↗