arXiv subjects

Xiaoyu Chen

Publications and source records attributed to Xiaoyu Chen.

At least 163 records · Page 9Linked to original sources

arXiv2019

Q-learning with UCB Exploration is Sample Efficient for Infinite-Horizon MDP

A fundamental question in reinforcement learning is whether model-free algorithms are sample efficient. Recently, Jin et al. \cite{jin2018q} proposed a Q-learning algorithm with UCB exploration policy, and proved it has nearly optimal regret bound for finite-horizon episodic MDP. In this paper, we adapt Q-learning with UCB-exploration bonus to infinite-horizon MDP with discounted rewards \emph{without} accessing a generative model. We show that the \textit{sample complexity of exploration} of our algorithm is bounded by $\tilde{O}({\frac{SA}{ε^2(1-γ)^7}})$. This improves the previously best known result of $\tilde{O}({\frac{SA}{ε^4(1-γ)^8}})$ in this setting achieved by delayed Q-learning \cite{strehl2006pac}, and matches the lower bound in terms of $ε$ as well as $S$ and $A$ except for logarithmic factors.

cs.LG↗

arXiv2019

Multiband superconductivity and possible nodal gap in RbCr$_{3}$As$_{3}$ revealed by Andreev reflection and single-particle tunneling measurements

By measuring point-contact Andreev reflection (PCAR) spectra in the newly discovered chromium-based quasi-one-dimensional superconductor RbCr$_{3}$As$_{3}$, we find clear evidence of two superconducting components, i.e., one having a gap value of about 1.8 meV and another with a gap value of about 5 meV. Since the current injection may have components in both the direction parallel and perpendicular to the [(Cr$_{3}$As$_{3}$)$^{-}$]$_\infty$ chains in the PCAR measurements, it naturally explains the two-component feature observed in this multi-band superconductor. Detailed analysis shows that the larger gap may have an $s$-wave nature. We then carry out the single-particle tunneling measurements based on a scanning tunneling spectroscope by using the needle-like sample as the tip, and in this case the measured current is mainly parallel to the [(Cr$_{3}$As$_{3}$)$^{-}$]$_\infty$ chains. The single-particle tunneling spectra show only one gap feature with a gap value of about 1.8 meV. Fitting to the single particle tunneling spectra indicates that the gap should have a large anisotropy or even node(s). We argue that the absence of the larger gap may be related to the direction of the injecting current. Therefore, our combined experiments show the multiband superconductivity with one gap being nodal or highly anisotropic.

cond-mat.supr-con↗

arXiv2019

Observation and characterization of the zero energy conductance peak in the vortex core state of FeTe$_{0.55}$Se$_{0.45}$

The search for the Majorana fermions in condensed matter physics has attracted much attention, partially because they may be used for the fault-tolerant quantum computation. It has been predicted that the Majorana zero mode may exist in the vortex core of topological superconductors. Recently, many iron-based superconductors are claimed to exhibit a topologically nontrivial surface state, including Fe(Te,Se). Some previous experiments through scanning tunneling microscopy (STM) have found zero-bias conductance peaks (ZBCP) within the vortex cores of Fe(Te,Se). However, our early experimental results have revealed the Caroli-de Gennes-Matricon (CdGM) discrete quantum levels in about 20% vortices. In many other vortices, we observed a dominant peak locating near zero bias. Here we show further study on the vortex core state of many more vortices in FeTe$_{0.55}$Se$_{0.45}$. In some vortices, if we take a certain criterion of bias voltage window near zero energy, we indeed see a zero energy mode. Some vortices exhibit zero energy bound state peaks with relatively symmetric background, which cannot be interpreted as the CdGM discrete states. The probability for vortices showing the ZBCP lowers down with the increase of magnetic field. Meanwhile it seems that the presence and absence of the ZBCP has no clear relationship with the Te/Se ratio on the surface. Temperature dependence of the spectra reveals that the ZBCP becomes weakened with increasing temperature and disappears at about 4 K. Our results provide a confirmed supplementary to the early claimed zero energy modes within the vortex cores of Fe(Te,Se). Detailed characterization of these zero energy modes versus magnetic field, temperature and spatial distribution of Te/Se will help to clarify its origin.

cond-mat.supr-con↗

arXiv2019

Non-destructive three-dimensional measurement of hand vein based on self-supervised network

At present, supervised stereo methods based on deep neural network have achieved impressive results. However, in some scenarios, accurate three-dimensional labels are inaccessible for supervised training. In this paper, a self-supervised network is proposed for binocular disparity matching (SDMNet), which computes dense disparity maps from stereo image pairs without disparity labels: In the self-supervised training, we match the stereo images densely to approximate the disparity maps and use them to warp the left and right images to estimate the right and left images; we build the loss function between estimated images and original images for self-supervised training, which adopts perceptual loss to help improve the quality of disparity maps in both detail and structure. Then, we use SDMNet to obtain disparities of hand vein. SDMNet has achieved excellent results on KITTI 2012, KITTI 2015, simulated vein dataset and real vein dataset, outperforming many state-of-the-art supervised matching methods.

cs.CV↗

arXiv2019

Distributed Bandit Learning: Near-Optimal Regret with Efficient Communication

We study the problem of regret minimization for distributed bandits learning, in which $M$ agents work collaboratively to minimize their total regret under the coordination of a central server. Our goal is to design communication protocols with near-optimal regret and little communication cost, which is measured by the total amount of transmitted data. For distributed multi-armed bandits, we propose a protocol with near-optimal regret and only $O(M\log(MK))$ communication cost, where $K$ is the number of arms. The communication cost is independent of the time horizon $T$, has only logarithmic dependence on the number of arms, and matches the lower bound except for a logarithmic factor. For distributed $d$-dimensional linear bandits, we propose a protocol that achieves near-optimal regret and has communication cost of order $\tilde{O}(Md)$, which has only logarithmic dependence on $T$.

cs.LG↗

arXiv2019

Zero-Energy Modes on Superconducting Bismuth Islands Deposited on Fe(Te,Se)

Topological superconductivity is one of the frontier research directions in condensed matter physics. One of the unique elementary excitations in topological superconducting state is the Majorana fermion (mode) which is its own antiparticle and obeys the non-Abelian statistics, and thus useful for constructing the fault-tolerant quantum computing. The evidence for Majorana fermions (mode) in condensed matter state is now quickly accumulated. Here we report the easily achievable zero-energy mode on the tunneling spectra on Bi islands deposited on the Fe(Te,Se) superconducting single crystals. We interpret this result as the consequence of proximity effect induced topological superconductivity on the Bi islands with strong spin-orbital coupling effect. The zero-energy mode is argued to be the signature of the Majorana modes in this size confined system.

cond-mat.supr-con↗

arXiv2019

The Permutation Module on Flag Varieties in Cross Characteristic

Let ${\bf G}$ be a connected reductive group over $\bar{\mathbb{F}}_q$, the algebraically closure of $\mathbb{F}_q$ (the finite field with $q=p^e$ elements), with the standard Frobenius map $F$. Let ${\bf B}$ be an $F$-stable Borel subgroup. Let $\Bbbk$ be a field of characteristic $r\neq p$. In this paper, we completely determine the composition factors of the induced module $Ind_{B}^{G}{tr}=\Bbbk{G}\otimes_{\Bbbk{\bf B}}$ tr (here $\Bbbk{H}$ is the group algebra of the group ${H}$, and tr is the trivial $B$-module). In particular, we find a new family of infinite dimensional irreducible abstract representations of $G$.

math.RT↗

arXiv2019

The Decomposition of Permutation Module for Infinite Chevalley Groups

Let ${\bf G}$ be a connected reductive group defined over $\mathbb{F}_q$, the finite field with $q$ elements. Let ${\bf B}$ be an Borel subgroup defined over $\mathbb{F}_q$. In this paper, we completely determine the composition factors of the induced module $\mathbb{M}(\op{tr})=\Bbbk{\bf G}\otimes_{\Bbbk{\bf B}}\op{tr}$ ($\op{tr}$ is the trivial ${\bf B}$-module) for any field $\Bbbk$.

math.RT↗

arXiv2019

Passive TCP Identification for Wired and WirelessNetworks: A Long-Short Term Memory Approach

Transmission control protocol (TCP) congestion control is one of the key techniques to improve network performance. TCP congestion control algorithm identification (TCP identification) can be used to significantly improve network efficiency. Existing TCP identification methods can only be applied to limited number of TCP congestion control algorithms and focus on wired networks. In this paper, we proposed a machine learning based passive TCP identification method for wired and wireless networks. After comparing among three typical machine learning models, we concluded that the 4-layers Long Short Term Memory (LSTM) model achieves the best identification accuracy. Our approach achieves better than 98% accuracy in wired and wireless networks and works for newly proposed TCP congestion control algorithms.

cs.NI↗

arXiv2019

Residual Pyramid Learning for Single-Shot Semantic Segmentation

Pixel-level semantic segmentation is a challenging task with a huge amount of computation, especially if the size of input is large. In the segmentation model, apart from the feature extraction, the extra decoder structure is often employed to recover spatial information. In this paper, we put forward a method for single-shot segmentation in a feature residual pyramid network (RPNet), which learns the main and residuals of segmentation by decomposing the label at different levels of residual blocks. Specifically speaking, we use the residual features to learn the edges and details, and the identity features to learn the main part of targets. At testing time, the predicted residuals are used to enhance the details of the top-level prediction. Residual learning blocks split the network into several shallow sub-networks which facilitates the training of the RPNet. We then evaluate the proposed method and compare it with recent state-of-the-art methods on CamVid and Cityscapes. The proposed single-shot segmentation based on RPNet achieves impressive results with high efficiency on pixel-level segmentation.

cs.CV↗

arXiv2018

A Mixed Integer Linear Programming Model for Multi-Satellite Scheduling

We address the multi-satellite scheduling problem with limited observation capacities that arises from the need to observe a set of targets on the Earth's surface using imaging resources installed on a set of satellites. We define and analyze the conflict indicators of all available visible time windows of missions, as well as the feasible time intervals of resources. The problem is then formulated as a mixed integer linear programming model, in which constraints are derived from a careful analysis of the interdependency between feasible time intervals that are eligible for observations. We apply the proposed model to several different problem instances that reflect real-world situations. The computational results verify that our approach is effective for obtaining optimum solutions or solutions with a very good quality.

math.OC↗

arXiv2018

Direct visualization of sign reversal s^+- superconducting gaps in FeTe0.55Se0.45

In many unconventional superconductors, the pairing of electrons is driven by the repulsive interaction, which leads to the sign reversal of superconducting gaps along the Fermi surfaces (FS) or between them. However, to measure this sign change is not easy and straightforward. It is known that, in superconductors with sign reversal gaps, non-magnetic impurities can break Cooper pairs leading to the quasiparticle density of states in the superconducting state. The standing waves of these quasiparticles will interfere each other leading to the quasiparticle interference (QPI) pattern which carries the phase message reflecting also the superconducting gap structure. Based on the recently proposed defect-bound-state QPI technique, we explore the applicability of this technique to a typical iron based superconductor FeTe$_{0.55}$Se$_{0.45}$ with roughly equivalent gap values on the electron and hole pockets connected by the wave vector q_2=(0,π). It is found that, on the negative energy side, with the energy slightly below the gap value, the phase reference quantity $|g(q,-E)|\cos(θ_{q,+E}-θ_{q,-E}) becomes negative and the amplitude is strongly enhanced with the scattering vector q_2, but that corresponding to the scattering between the electron-electron pockets, namely q_3=(π,π), keeps all positive. This is well consistent with the theoretical expectation of the s^+- pairing gap and thus serves as a direct visualization of the sign reversal gaps. This experimental observation is also supported by the theoretical calculations with the Fermi surface structure and s^+- pairing gap.

cond-mat.supr-con↗

arXiv2018

Discrete Potts Model for Generating Superpixels on Noisy Images

Many computer vision applications, such as object recognition and segmentation, increasingly build on superpixels. However, there have been so far few superpixel algorithms that systematically deal with noisy images. We propose to first decompose the image into equal-sized rectangular patches, which also sets the maximum superpixel size. Within each patch, a Potts model for simultaneous segmentation and denoising is applied, that guarantees connected and non-overlapping superpixels and also produces a denoised image. The corresponding optimization problem is formulated as a mixed integer linear program (MILP), and solved by a commercial solver. Extensive experiments on the BSDS500 dataset images with noises are compared with other state-of-the-art superpixel methods. Our method achieves the best result in terms of a combined score (OP) composed of the under-segmentation error, boundary recall and compactness.

cs.CV↗

arXiv2018

On the principal series representations of semisimple groups with Frobenius maps

Let ${\bf G}$ be a simply connected semisimple algebraic group over $\Bbbk=\bar{\mathbb{F}}_q$, the algebraically closure of $\mathbb{F}_q$ (the finite field with $q=p^e$ elements), and $F$ be the standard Frobenius map. Let ${\bf B}$ be an $F$-stable Borel subgroup and ${\bf T}$ an $F$-stable maximal torus contained in ${\bf B}$. Set $G_{q^r}={\bf G}^{F^r}$ and $B_{q^r}={\bf B}^{F^r}$ for any $r>0$. This paper studies the original induced module $\op{Ind}_{\bf B}^{\bf G}λ=\Bbbk{\bf G}\otimes_{\Bbbk{\bf B}}λ$ (here $\Bbbk{\bf H}$ is the group algebra of the group ${\bf H}$, and $λ$ is a rational character of ${\bf T}$ regarded as a ${\bf B}$-module). We show that if $λ$ is regular and dominant, then there is a surjective ${\bf G}$-module homomorphism $\op{Ind}_{B_{q^r}}^{\bf G}λ\rightarrow \op{St}\otimes L(λ)$ for any $r>0$, where $\op{St}$ is the infinite dimensional Steinberg module defined by Nanhua Xi. As a consequence, we show that $\op{Ind}_{\bf B}^{\bf G}λ$ is irreducible if $λ$ if and only if $λ$ is regular and antidominant. Moreover, for ${\bf G}=SL_2(\bar{\mathbb{F}}_q)$ and $0<λ<p$, we show that $\op{Ind}_{\bf B}^{\bf G}λ$ have infinite many composition factors with each finite dimensional. Consequently, we find certain $λ$ for which $\op{Ind}_{\bf B}^{\bf G}λ$ has an infinite submodule filtration for the general ${\bf G}$.

math.RT↗

arXiv2018

Some Non Quasi-finite irreducible Modules of Semisimple Groups with Frobenius Maps

This paper is the continuation of \cite{CXY}. Let ${\bf G}$ be a simply connected semisimple algebraic group over $\Bbbk=\bar{\mathbb{F}}_q$, the algebraically closure of $\mathbb{F}_q$ (the finite field with $q=p^e$ elements), and $F$ be the standard Frobenius map. Let ${\bf B}$ be an $F$-stable Borel subgroup and ${\bf T}$ an $F$-stable maximal torus contained in ${\bf B}$. This paper studies the original induced module $\op{Ind}_{\bf B}^{\bf G}λ=\Bbbk{\bf G}\otimes_{\Bbbk{\bf B}}λ$ (here $\Bbbk{\bf H}$ is the group algebra of the group ${\bf H}$, and $λ$ is a rational character of ${\bf T}$ regarded as a ${\bf B}$-module). We show that if $λ$ is antidominant and not trivial, then certain submodule of $\op{Ind}_{\bf B}^{\bf G}λ$ is irreducible and non quasi-finite.

math.RT↗

arXiv2017

Discrete energy levels of Caroli-de Gennes-Martricon states in quantum limit due to small Fermi energy in FeTe$_{0.55}$Se$_{0.45}$

Caroli-de Gennes-Martricon (CdGM) states were predicted in 1964 as low energy excitations within vortex cores of type-II superconductors. In the quantum limit, namely $T/T_\mathrm{c} \ll Δ/E_\mathrm{F}$, the energy levels of these states were predicted to be discrete with the basic levels at $E_μ= \pm μΔ^2/E_\mathrm{F}$ ($μ= 1/2$, $3/2$, $5/2$, ...). However, due to the small ratio of $Δ/E_\mathrm{F}$ in most type-II superconductors, it is very difficult to observe the discrete CdGM states, but rather a symmetric peak appears at zero-bias at the vortex center. Here we report the clear observation of these discrete energy levels of CdGM states in FeTe$_{0.55}$Se$_{0.45}$. The rather stable energies of these states versus space clearly validates our conclusion. Analysis based on the energies of these CdGM states indicates that the Fermi energy in the present system is very small.

cond-mat.supr-con↗

arXiv2017

Towards Near Real-Time BGP Deep Analysis: A Big-Data Approach

BGP (Border Gateway Protocol) serves as the primary routing protocol for the Internet, enabling Autonomous Systems (individual network operators) to exchange network reachability information. Alongside significant on-going research and development efforts, there is a practical need to understand the nature of events that occur on the Internet. Network operators are acutely aware of security-related incidents such as 'Prefix Hijacking' as well as the impact of network instabilities that ripple through the Internet. Recent research focused on the study of BGP anomalies (both network/prefix instability and security-related incidents) has been based on the analysis of historical logs. Further analysis to understand the nature of these anomalous events is not always sufficient to be able to differentiate malicious activities, such as prefix- or sub-prefix- hijacking, from those events caused by inadvertent misconfigurations. In addition, such techniques are challenged by a lack of sufficient resources to store and process data feeds in real-time from multiple BGP Vantage Points (VPs). In this paper, we present a BGP Deep-analysis application developed using the PNDA (Platform for Network Data Analytics) 'Big-Data' platform. PNDA provides a highly scalable environment that enables the ingestion and processing of 'live' BGP feeds from many vantage points in a schema-agnostic manner. The Apache Spark-based application, in conjunction with PNDA's distributed processing capabilities, is able to perform high-level insights as well as near-to-real-time statistical analysis

cs.NI↗

arXiv2016

On High-Order Capacity Statistics of Spectrum Aggregation Systems over $κ$-$μ$ and $κ$-$μ$ shadowed Fading Channels

The frequency scarcity imposed by fast growing demand for mobile data service requires promising spectrum aggregation systems. The so-called higher-order statistics (HOS) of the channel capacity is a suitable metric on the system performance. While prior relevant works have improved our knowledge on the HOS characterization of spectrum aggregation systems, an analytical framework encompassing generalized fading models of interest is not yet available. In this paper, we pursue a detailed HOS analysis of $κ$-$μ$ and $κ$-$μ$ shadowed fading channels by deriving novel and exact expressions. Furthermore, the simplified HOS expressions for the asymptotically low and high signal-to-noise regimes are derived. Several important statistical measures, such as amount of fading, amount of dispersion, reliability, skewness, and kurtosis, are obtained by using the HOS results. More importantly, the useful implications of system and fading parameters on spectrum aggregation systems are investigated for channel selection. Finally, all derived expressions are validated via Monte-Carlo simulations.

cs.IT↗