Search arXiv⌕ Search

arXiv subjects

Peng Liu

Publications and source records attributed to Peng Liu.

At least 433 records · Page 24Linked to original sources
arXiv2018

Extremes of vector-valued Gaussian processes with Trend

Let $X(t)=(X_1(t), \dots, X_n(t)), t\in \mathcal{T}\subset \mathbb{R} $ be a centered vector-valued Gaussian process with independent components and continuous trajectories, and $h(t)=(h_1(t),\dots, h_n(t)), t\in \mathcal{T} $ be a vector-valued continuous function. We investigate the asymptotics of $$\mathbb{P}\left(\sup_{t\in \mathcal{T} } \min_{1\leq i\leq n}(X_i(t)+h_i(t))>u\right)$$ as $u\to\infty$. As an illustration to the derived results we analyze two important classes of $X(t)$: with locally-stationary structure and with varying variances of the coordinates, and calculate exact asymptotics of simultaneous ruin probability and ruin time in a Gaussian risk model.

math.PR↗
arXiv2018

The time of ultimate recovery in Gaussian risk model

We analyze the distance $\mathcal{R}_T(u)$ between the first and the last passage time of $\{X(t)-ct:t\in [0,T]\}$ at level $u$ in time horizon $T\in(0,\infty]$, where $X$ is a centered Gaussian process with stationary increments and $c\in\mathbb{R}$, given that the first passage time occurred before $T$. Under some tractable assumptions on $X$, we find $Δ(u)$ and $G(x)$ such that $$\lim_{u\to\infty}\mathbb{P}\left(\mathcal{R}_T(u)>Δ(u)x\right)=G(x),$$ for $x\geq 0$. We distinguish two scenarios: $T<\infty$ and $T=\infty$, that lead to qualitatively different asymptotics. The obtained results provide exact asymptotics of the ultimate recovery time after the ruin in Gaussian risk model.

math.PR↗
arXiv2018

Massive neutron star with strangeness in a relativistic mean-field model with a high-density cut-off

The properties of strangeness neutron star are studied within relativistic mean-field (RMF) model via including a logarithmic interaction as a function of scalar meson field. This logarithmic interaction, named as the $σ$-cut potential, can largely reduce the attractive contributions of scalar meson field at high density without any influence on nuclear structure around normal saturation density. In this work, the TM1 parameter set is chosen as the RMF interaction, while the strengths of logarithmic interaction are constrained by the properties of finite nuclei so that we can obtain a reasonable effective nucleon-nucleon interaction. The hyperons, $Λ,~Σ$, and $Ξ$ are also considered in neutron stars within this framework, whose coupling constants with mesons are determined by the latest hyperon-nucleon and $Λ$-$Λ$ potentials extracted from the experimental data of hypernuclei. The maximum mass of neutron star can be larger than two solar mass with these hyperons. Furthermore, the nucleon mass at high density will be saturated due to this additional $σ$-cut potential, which is consistent with the conclusions from the microscopic calculations such as, Brueckner-Hartree-Fock theory and quark mean-field model.

nucl-th↗
arXiv2017

On convergence rates of game theoretic reinforcement learning algorithms

This paper investigates a class of multi-player discrete games where each player aims to maximize its own utility function. Each player does not know the other players' action sets, their deployed actions or the structures of its own or the others' utility functions. Instead, each player only knows its own deployed actions and its received utility values in recent history. We propose a reinforcement learning algorithm which converges to the set of action profiles which have maximal stochastic potential with probability one. Furthermore, the convergence rate of the proposed algorithm is quantified. The algorithm performance is verified using two case studies in the smart grid and cybersecurity.

math.OC↗
arXiv2017

Extremes of $γ$-reflected Gaussian process with stationary increments

For a given centered Gaussian process with stationary increments $\{X(t), t\geq 0\}$ and $c>0$, let $$ W_γ(t)=X(t)-ct-γ\inf_{0\leq s\leq t}\left(X(s)-cs\right), \quad t\geq 0$$ denote the $γ$-reflected process, where $γ\in (0,1)$. This process is introduced in the context of risk theory to model surplus process that include tax payments of loss-carry forward type.In this contribution we derive asymptotic approximations of both the ruin probability and the joint distribution of first and last passage times given that ruin occurs. We apply our findings to the cases with $X$ being the multiplex fractional Brownian motion and the integrated Gaussian processes. As a by-product we derive an extension of Piterbarg inequality \KD{for} threshold-dependent random fields.

math.PR↗
arXiv2017

Holographic Butterfly Effect at Quantum Critical Points

When the Lyapunov exponent $λ_L$ in a quantum chaotic system saturates the bound $λ_L\leqslant 2πk_BT$, it is proposed that this system has a holographic dual described by a gravity theory. In particular, the butterfly effect as a prominent phenomenon of chaos can ubiquitously exist in a black hole system characterized by a shockwave solution near the horizon. In this paper we propose that the butterfly velocity can be used to diagnose quantum phase transition (QPT) in holographic theories. We provide evidences for this proposal with an anisotropic holographic model exhibiting metal-insulator transitions (MIT), in which the derivatives of the butterfly velocity with respect to system parameters characterizes quantum critical points (QCP) with local extremes in zero temperature limit. We also point out that this proposal can be tested by experiments in the light of recent progress on the measurement of out-of-time-order correlation function (OTOC).

hep-th↗
arXiv2017

Holographic superconductivity from higher derivative theory

We construct a $6$ derivative holographic superconductor model in the $4$-dimensional bulk spacetimes, in which the normal state describes a quantum critical (QC) phase. The phase diagram $(γ_1,\hat{T}_c)$ and the condensation as the function of temperature are worked out numerically. We observe that with the decrease of the coupling parameter $γ_1$, the critical temperature $\hat{T}_c$ decreases and the formation of charged scalar hair becomes harder. We also calculate the optical conductivity. An appealing characteristic is a wider extension of the superconducting energy gap, comparing with that of $4$ derivative theory. It is expected that this phenomena can be observed in the real materials of high temperature superconductor. Also the Homes' law in our present models with $4$ and $6$ derivative corrections is explored. We find that in certain range of parameters $γ$ and $γ_1$, the experimentally measured value of the universal constant $C$ in Homes' law can be obtained.

hep-th↗
arXiv2017

Pure Even Harmonic Generation from Oriented CO in Linearly Polarized Laser Fields

The first high harmonic spectrum, containing only the odd orders, was observed in experiments 30 years ago (1987). However, a spectrum containing pure even harmonics has never been observed. We investigate the generation of pure even harmonics from oriented CO molecules in linearly polarized laser fields employing the time-dependent density-functional theory. We find that the even harmonics, with no odd orders, are generated with the polarization perpendicular to the laser polarization, when the molecular axis of CO is perpendicular to the laser polarization. Generation of pure even harmonics reveals a type of dipole acceleration originating from the permanent dipole moment. This phenomenon exists in all system with permanent dipole moments, including bulk crystal and polyatomic molecules.

physics.optics↗
arXiv2017

GaDei: On Scale-up Training As A Service For Deep Learning

Deep learning (DL) training-as-a-service (TaaS) is an important emerging industrial workload. The unique challenge of TaaS is that it must satisfy a wide range of customers who have no experience and resources to tune DL hyper-parameters, and meticulous tuning for each user's dataset is prohibitively expensive. Therefore, TaaS hyper-parameters must be fixed with values that are applicable to all users. IBM Watson Natural Language Classifier (NLC) service, the most popular IBM cognitive service used by thousands of enterprise-level clients around the globe, is a typical TaaS service. By evaluating the NLC workloads, we show that only the conservative hyper-parameter setup (e.g., small mini-batch size and small learning rate) can guarantee acceptable model accuracy for a wide range of customers. We further justify theoretically why such a setup guarantees better model convergence in general. Unfortunately, the small mini-batch size causes a high volume of communication traffic in a parameter-server based system. We characterize the high communication bandwidth requirement of TaaS using representative industrial deep learning workloads and demonstrate that none of the state-of-the-art scale-up or scale-out solutions can satisfy such a requirement. We then present GaDei, an optimized shared-memory based scale-up parameter server design. We prove that the designed protocol is deadlock-free and it processes each gradient exactly once. Our implementation is evaluated on both commercial benchmarks and public benchmarks to demonstrate that it significantly outperforms the state-of-the-art parameter-server based implementation while maintaining the required accuracy and our implementation reaches near the best possible runtime performance, constrained only by the hardware limitation. Furthermore, to the best of our knowledge, GaDei is the only scale-up DL system that provides fault-tolerance.

stat.ML↗
arXiv2017

Introduction of Improved Repairing Locality into Secret Sharing Schemes with Perfect Security

Repairing locality is an appreciated feature for distributed storage, in which a damaged or lost data share can be repaired by accessing a subset of other shares much smaller than is required for decoding the complete data. However for Secret Sharing (SS) schemes, it has been proven theoretically that local repairing can not be achieved with perfect security for the majority of threshold SS schemes, where all the shares are equally regarded in both secret recovering and share repairing. In this paper we make an attempt on decoupling the two processes to make secure local repairing possible. Dedicated repairing redundancies only for the repairing process are generated, which are random numbers to the original secret. Through this manner a threshold SS scheme with improved repairing locality is achieved on the condition that security of repairing redundancies is ensured, or else our scheme degenerates into a perfect access structure that is equivalent to the best existing schemes can do. To maximize security of the repairing redundancies, a random placement mechanism is also proposed.

cs.CR↗
arXiv2017

Exploiting Physical Dynamics to Detect Actuator and Sensor Attacks in Mobile Robots

Mobile robots are cyber-physical systems where the cyberspace and the physical world are strongly coupled. Attacks against mobile robots can transcend cyber defenses and escalate into disastrous consequences in the physical world. In this paper, we focus on the detection of active attacks that are capable of directly influencing robot mission operation. Through leveraging physical dynamics of mobile robots, we develop RIDS, a novel robot intrusion detection system that can detect actuator attacks as well as sensor attacks for nonlinear mobile robots subject to stochastic noises. We implement and evaluate a RIDS on Khepera mobile robot against concrete attack scenarios via various attack channels including signal interference, sensor spoofing, logic bomb, and physical damage. Evaluation of 20 experiments shows that the averages of false positive rates and false negative rates are both below 1%. Average detection delay for each attack remains within 0.40s.

cs.CR↗
arXiv2017

Learning Pixel-Distribution Prior with Wider Convolution for Image Denoising

In this work, we explore an innovative strategy for image denoising by using convolutional neural networks (CNN) to learn pixel-distribution from noisy data. By increasing CNN's width with large reception fields and more channels in each layer, CNNs can reveal the ability to learn pixel-distribution, which is a prior existing in many different types of noise. The key to our approach is a discovery that wider CNNs tends to learn the pixel-distribution features, which provides the probability of that inference-mapping primarily relies on the priors instead of deeper CNNs with more stacked nonlinear layers. We evaluate our work: Wide inference Networks (WIN) on additive white Gaussian noise (AWGN) and demonstrate that by learning the pixel-distribution in images, WIN-based network consistently achieves significantly better performance than current state-of-the-art deep CNN-based methods in both quantitative and visual evaluations. \textit{Code and models are available at \url{https://github.com/cswin/WIN}}.

cs.CV↗
arXiv2017

TrustShadow: Secure Execution of Unmodified Applications with ARM TrustZone

The rapid evolution of Internet-of-Things (IoT) technologies has led to an emerging need to make it smarter. A variety of applications now run simultaneously on an ARM-based processor. For example, devices on the edge of the Internet are provided with higher horsepower to be entrusted with storing, processing and analyzing data collected from IoT devices. This significantly improves efficiency and reduces the amount of data that needs to be transported to the cloud for data processing, analysis and storage. However, commodity OSes are prone to compromise. Once they are exploited, attackers can access the data on these devices. Since the data stored and processed on the devices can be sensitive, left untackled, this is particularly disconcerting. In this paper, we propose a new system, TrustShadow that shields legacy applications from untrusted OSes. TrustShadow takes advantage of ARM TrustZone technology and partitions resources into the secure and normal worlds. In the secure world, TrustShadow constructs a trusted execution environment for security-critical applications. This trusted environment is maintained by a lightweight runtime system that coordinates the communication between applications and the ordinary OS running in the normal world. The runtime system does not provide system services itself. Rather, it forwards requests for system services to the ordinary OS, and verifies the correctness of the responses. To demonstrate the efficiency of this design, we prototyped TrustShadow on a real chip board with ARM TrustZone support, and evaluated its performance using both microbenchmarks and real-world applications. We showed TrustShadow introduces only negligible overhead to real-world applications.

cs.CR↗
arXiv2017

Hey, you, keep away from my device: remotely implanting a virus expeller to defeat Mirai on IoT devices

Mirai is botnet which targets out-of-date Internet-of-Things (IoT) devices. The disruptive Distributed Denial of Service (DDoS) attack last year has hit major Internet companies, causing intermittent service for millions of Internet users. Since the affected devices typically do not support firmware update, it becomes challenging to expel these vulnerable devices in the wild. Both industry and academia have made great efforts in amending the situation. However, none of these efforts is simple to deploy, and at the same time effective in solving the problem. In this work, we design a collaborative defense strategy to tackle Mirai. Our key idea is to take advantage of human involvement in the least aggressive way. In particular, at a negotiated time slot, a customer is required to reboot the compromised device, then a "white" Mirai operated by the manufacturer breaks into the clean-state IoT devices immediately. The "white" Mirai expels other malicious Mirai variants, blocks vulnerable ports, and keeps a heart-beat connection with the server operated by the manufacturer. Once the heart-beat is lost, the server re-implants the "white" Mirai instantly. We have implemented a full prototype of the designed system, and the results show that our system can evade Mirai attacks effectively.

cs.CR↗
arXiv2017

Uniform tail approximation of homogenous functionals of Gaussian fields

Let $X(t),t\in R^d$ be a centered Gaussian random field with continuous trajectories and set $ξ_u(t)= X(f(u)t),t\in R^d$ with $f$ some positive function. Classical results establish the tail asymptotics of $P\{ Γ(ξ_u) > u\}$ as $u\to \infty$ with $Γ(ξ_u)= \sup_{t \in [0,T ]^d} ξ_u(t),T>0$ by requiring that $f(u) \to 0$ with speed controlled by the local behaviour of the correlation function of $X$. Recent research shows that for applications more general continuous functionals than supremum should be considered and the Gaussian field can depend also on some additional parameter $τ_u \in K$, say $ξ_{u,τ_u}(t),t\in R^d$. In this contribution we derive uniform approximations of $P\{ Γ(ξ_{u,τ_u})> u\}$ with respect to $τ_u$ in some index set $K_u$, as $u\to\infty$. Our main result have important theoretical implications; two applications are already included in [10,11]. In this paper we present three additional ones, namely i) we derive uniform upper bounds for the probability of double-maxima, ii) we extend Piterbarg-Prisyazhnyuk theorem to some large classes of homogeneous functionals of centered Gaussian fields $ξ_{u}$, and iii) we show the finiteness of generalized Piterbarg constants.

math.PR↗
arXiv2017

Distributed Model Predictive Control of Spatially Interconnected Systems Using Switched Cost Functions

This note proposes a distributed model predictive control (DMPC) scheme with switched cost functions for a class of spatially interconnected systems with communication constraints. Non-iterative and parallel communication strategy is considered to ensure that all distributed controllers complete input updates at each single information exchange step. The proposed DMPC scheme switches the optimization index on a switching surface generated by control invariant sets. With the index-switching strategy, stability of the origin is ensured by a terminal control law. Convergence conditions of the optimal cost to zero are established taking into account the causal link between the presumed trajectory and the optimized trajectory of the previous step. The compatibility constraints preserve the quadratic program property that is desired in practical applications. It is also observed that the proposed DMPC scheme has benefits dealing with systems that need to take into account safety-related spatial constraints.

math.OC↗
arXiv2017

From Physical to Cyber: Escalating Protection for Personalized Auto Insurance

Nowadays, auto insurance companies set personalized insurance rate based on data gathered directly from their customers' cars. In this paper, we show such a personalized insurance mechanism -- wildly adopted by many auto insurance companies -- is vulnerable to exploit. In particular, we demonstrate that an adversary can leverage off-the-shelf hardware to manipulate the data to the device that collects drivers' habits for insurance rate customization and obtain a fraudulent insurance discount. In response to this type of attack, we also propose a defense mechanism that escalates the protection for insurers' data collection. The main idea of this mechanism is to augment the insurer's data collection device with the ability to gather unforgeable data acquired from the physical world, and then leverage these data to identify manipulated data points. Our defense mechanism leveraged a statistical model built on unmanipulated data and is robust to manipulation methods that are not foreseen previously. We have implemented this defense mechanism as a proof-of-concept prototype and tested its effectiveness in the real world. Our evaluation shows that our defense mechanism exhibits a false positive rate of 0.032 and a false negative rate of 0.013.

cs.CR↗