Searcharxiv⌕ Search

arXiv subjects

Tao Chen

Publications and source records attributed to Tao Chen.

557 records · Page 31Linked to original sources
arXiv2015

Online QoS Modeling in the Cloud: A Hybrid and Adaptive Multi-Learners Approach

Given the on-demand nature of cloud computing, managing cloud-based services requires accurate modeling for the correlation between their Quality of Service (QoS) and cloud configurations/resources. The resulted models need to cope with the dynamic fluctuation of QoS sensitivity and interference. However, existing QoS modeling in the cloud are limited in terms of both accuracy and applicability due to their static and semi- dynamic nature. In this paper, we present a fully dynamic multi- learners approach for automated and online QoS modeling in the cloud. We contribute to a hybrid learners solution, which improves accuracy while keeping model complexity adequate. To determine the inputs of QoS model at runtime, we partition the inputs space into two sub-spaces, each of which applies different symmetric uncertainty based selection techniques, and we then combine the sub-spaces results. The learners are also adaptive; they simultaneously allow several machine learning algorithms to model QoS function and dynamically select the best model for prediction on the fly. We experimentally evaluate our models using RUBiS benchmark and realistic FIFA 98 workload. The results show that our multi-learners approach is more accurate and effective in contrast to the other state-of-the-art approaches.

cs.DC↗
arXiv2014

Dynamic Conic Finance via Backward Stochastic Difference Equations

We present an arbitrage free theoretical framework for modeling bid and ask prices of dividend paying securities in a discrete time setup using theory of dynamic acceptability indices. In the first part of the paper we develop the theory of dynamic subscale invariant performance measures, on a general probability space, and discrete time setup. We prove a representation theorem of such measures in terms of a family of dynamic convex risk measures, and provide a representation of dynamic risk measures in terms of g-expectations, and solutions of BS$Δ$Es with convex drivers. We study the existence and uniqueness of the solutions, and derive a comparison theorem for corresponding BS$Δ$Es. In the second part of the paper we discuss a market model for dividend paying securities by introducing the pricing operators that are defined in terms of dynamic acceptability indices, and find various properties of these operators. Using these pricing operators, we define the bid and ask prices for the underlying securities and then for derivatives in this market. We show that the obtained market model is arbitrage free, and we also prove a series of properties of these prices.

q-fin.PR↗
arXiv2014

DeepSentiBank: Visual Sentiment Concept Classification with Deep Convolutional Neural Networks

This paper introduces a visual sentiment concept classification method based on deep convolutional neural networks (CNNs). The visual sentiment concepts are adjective noun pairs (ANPs) automatically discovered from the tags of web photos, and can be utilized as effective statistical cues for detecting emotions depicted in the images. Nearly one million Flickr images tagged with these ANPs are downloaded to train the classifiers of the concepts. We adopt the popular model of deep convolutional neural networks which recently shows great performance improvement on classifying large-scale web-based image dataset such as ImageNet. Our deep CNNs model is trained based on Caffe, a newly developed deep learning framework. To deal with the biased training data which only contains images with strong sentiment and to prevent overfitting, we initialize the model with the model weights trained from ImageNet. Performance evaluation shows the newly trained deep CNNs model SentiBank 2.0 (or called DeepSentiBank) is significantly improved in both annotation accuracy and retrieval performance, compared to its predecessors which mainly use binary SVM classification models.

cs.CV↗
arXiv2014

Latent Tree Models and Approximate Inference in Bayesian Networks

We propose a novel method for approximate inference in Bayesian networks (BNs). The idea is to sample data from a BN, learn a latent tree model (LTM) from the data offline, and when online, make inference with the LTM instead of the original BN. Because LTMs are tree-structured, inference takes linear time. In the meantime, they can represent complex relationship among leaf nodes and hence the approximation accuracy is often good. Empirical evidence shows that our method can achieve good approximation accuracy at low online computational cost.

cs.LG↗
arXiv2013

Bounded Geometry and Families of Meromorphic Functions with Two Asymptotic Values

In this paper we consider the topological class, denoted by AV2, of universal covering maps from the plane to the sphere with two removed points. We prove that an element f in AV2 with finite post-singular set is combinatorially equivalent to a meromorphic transcendental map g with constant Schwarzian derivative if any only if f satisfies a Bounded Geometry condition.

math.DS↗
arXiv2013

Bounded Geometry and Characterization of post-singularly Finite $(p,q)$-Exponential Maps

In this paper we define a topological class of branched covering maps of the plane called {\em topological exponential maps of type $(p,q)$} and denoted by $\TE_{p,q}$, where $p\geq 0$ and $q\geq 1$. We follow the framework given in \cite{Ji} to study the problem of combinatorially characterizing an entire map $P e^{Q}$, where $P$ is a polynomial of degree $p$ and $Q$ is a polynomial of degree $q$ using an {\em iteration scheme defined by Thurston} and a {\em bounded geometry condition}. We first show that an element $f \in {\TE}_{p,q}$ with finite post-singular set is combinatorially equivalent to an entire map $P e^{Q}$ if and only if it has bounded geometry with compactness. Thus to complete the characterization, we only need to check that the bounded geometry actually implies compactness. We show this for some $f\in \TE_{p,1}$, $p\geq 1$. Our main result in this paper is that a post-singularly finite map $f$ in $\TE_{0,1}$ or a post-singularly finite map $f$ in $\TE_{p,1}$, $p\geq 1$, with only one non-zero simple breanch point $c$ such that either $c$ is periodic or $c$ and $f(c)$ are both not periodic, is combinatorially equivalent to a post-singularly finite entire map of either the form $e^{λz}$ or the form $ αz^{p}e^{λz}$, where $α=(-λ/p)^{p}e^{- λ(-p/λ)^{p}}$, respectively, if and only if it has bounded geometry. This is the first result in this direction for a family of transcendental holomorphic maps with critical points.

math.DS↗
arXiv2013

On Improving Energy Efficiency within Green Femtocell Networks: A Hierarchical Reinforcement Learning Approach

One of the efficient solutions of improving coverage and increasing capacity in cellular networks is the deployment of femtocells. As the cellular networks are becoming more complex, energy consumption of whole network infrastructure is becoming important in terms of both operational costs and environmental impacts. This paper investigates energy efficiency of two-tier femtocell networks through combining game theory and stochastic learning. With the Stackelberg game formulation, a hierarchical reinforcement learning framework is applied for studying the joint expected utility maximization of macrocells and femtocells subject to the minimum signal-to-interference-plus-noise-ratio requirements. In the learning procedure, the macrocells act as leaders and the femtocells are followers. At each time step, the leaders commit to dynamic strategies based on the best responses of the followers, while the followers compete against each other with no further information but the leaders' transmission parameters. In this paper, we propose two reinforcement learning based intelligent algorithms to schedule each cell's stochastic power levels. Numerical experiments are presented to validate the investigations. The results show that the two learning algorithms substantially improve the energy efficiency of the femtocell networks.

cs.LG↗
arXiv2012

Improving Energy Efficiency in Femtocell Networks: A Hierarchical Reinforcement Learning Framework

This paper investigates energy efficiency for two-tier femtocell networks through combining game theory and stochastic learning. With the Stackelberg game formulation, a hierarchical reinforcement learning framework is applied to study the joint average utility maximization of macrocells and femtocells subject to the minimum signal-to-interference-plus-noise-ratio requirements. The macrocells behave as the leaders and the femtocells are followers during the learning procedure. At each time step, the leaders commit to dynamic strategies based on the best responses of the followers, while the followers compete against each other with no further information but the leaders' strategy information. In this paper, we propose two learning algorithms to schedule each cell's stochastic power levels, leading by the macrocells. Numerical experiments are presented to validate the proposed studies and show that the two learning algorithms substantially improve the energy efficiency of the femtocell networks.

cs.LG↗
arXiv2011

Creating a Live, Public Short Message Service Corpus: The NUS SMS Corpus

Short Message Service (SMS) messages are largely sent directly from one person to another from their mobile phones. They represent a means of personal communication that is an important communicative artifact in our current digital era. As most existing studies have used private access to SMS corpora, comparative studies using the same raw SMS data has not been possible up to now. We describe our efforts to collect a public SMS corpus to address this problem. We use a battery of methodologies to collect the corpus, paying particular attention to privacy issues to address contributors' concerns. Our live project collects new SMS message submissions, checks their quality and adds the valid messages, releasing the resultant corpus as XML and as SQL dumps, along with corpus statistics, every month. We opportunistically collect as much metadata about the messages and their sender as possible, so as to enable different types of analyses. To date, we have collected about 60,000 messages, focusing on English and Mandarin Chinese.

cs.CL↗
arXiv2009

Cloud Networking Formation in CogMesh Environment

As radio spectrum usage paradigm moving from the traditional command and control allocation scheme to the open spectrum allocation scheme, wireless networks meet new opportunities and challenges. In this article we introduce the concept of cognitive wireless mesh (CogMesh) networks and address the unique problem in such a network. CogMesh is a self-organized distributed network architecture combining cognitive technologies with the mesh structure in order to provide a uniform service platform over a wide range of networks. It is based on dynamic spectrum access (DSA) and featured by self-organization, self-configuration and self-healing. The unique problem in CogMesh is the common control channel problem, which is caused by the opportunistic spectrum sharing nature of secondary users (SU) in the network. More precisely, since the channels of SUs are fluctuating according to the radio environment, it is difficult to find always available global common control channels. This puts a significant challenge on the network design. We develop the control cloud based control channel selection and cluster based network formation techniques to tackle this problem. Moreover, we show in this article that the swarm intelligence is a good candidate to deal with the control channel problem in CogMesh. Since the study of cognitive wireless networks (CWN) is still in its early phase, the ideas provided in this article act as a catalyst to inspire new solutions in this field.

cs.NI↗
arXiv2005

Choquet expectation and Peng's g-expectation

In this paper we consider two ways to generalize the mathematical expectation of a random variable, the Choquet expectation and Peng's g-expectation. An open question has been, after making suitable restrictions to the class of random variables acted on by the Choquet expectation, for what class of expectation do these two definitions coincide? In this paper we provide a necessary and sufficient condition which proves that the only expectation which lies in both classes is the traditional linear expectation. This settles another open question about whether Choquet expectation may be used to obtain Monte Carlo-like solution of nonlinear PDE: It cannot, except for some very special cases.

math.PR↗
arXiv1999

Transport properties of the hierarchical model for stretched polymers

We prove that the hierarchical fractal model recently proposed for describing the stretched polymers [A. N. Samukhin et al, Phys. Rev. Lett.{\bf 78}, 326(1997)] is equivalent to a one-dimensional chain with hierarchical aperiodic structure. By use of the transfer matrix technique we calculate the electronic transmission and the dc conductance. We find that there exist sharp-edged transmission subbands and gaps, but the transmission subbands are rich in substructures which show self-similarity. The temperature dependence of conductance $g (T)$ is sensitive to the variation of the Fermi level and to the structure parameters $m$ and $n$ of the original fractal structure. The relationship of the obtained results to the experimental data and other theoretical works is discussed.

cond-mat.dis-nn↗
arXiv1997

Helicity Modulus and Fluctuating Type II Superconductors: Elastic Approximation and Numerical Simulations

We develop the helicity modulus as a criterion for superconducting order in the mixed phase of a fluctuating type II superconductor. We show that there is a duality relation between this helicity modulus and the superfluid density of a system of analog 2D bosons. We show that the vortex line lattice exhibits a perfect Meissner effect with respect to a shearing perturbation of the applied magnetic field, and this becomes our creterion for "longitudinal superconductivity" parallel to the applied field. We present arguments based on the 2D boson analogy, as well as the results of numerical simulations, that suggest that longitudinal superconductivity can persist into the vortex line liquid state for systems of finite thickness, comparable to those commonly found in experiments.

cond-mat.supr-con↗
arXiv1996

Phase Transitions in a Model Anisotropic High Tc Superconductor

We carry out simulations of the anisotropic uniformly frustrated 3D XY model, as a model for vortex line fluctuations in high Tc superconductors. We compute the phase diagram as a function of temperature and anisotropy, for a fixed applied magnetic field. We find that superconducting coherence parallel to the field persists into the vortex line liquid state, and that this transition lies well below the "mean-field" cross-over from the vortex line liquid to the normal state.

cond-mat.supr-con↗
arXiv1993

Helicity Modulus and Meissner Effect in a Fluctuating Type II Superconductor

The helicity modulus for a fluctuating type II superconductor is computed within the elastic medium approximation, as a probe of superconducting phase coherence and the Meissner effect in the mixed state. We argue that at the vortex line lattice melting transition, there remains superconducting coherence parallel to the applied magnetic field, provided the vortex line liquid retains a finite shear modulus at finite wavevector.

cond-mat↗