arXiv subjects

John C. S. Lui

Publications and source records attributed to John C. S. Lui.

132 records · Page 8Linked to original sources

arXiv2012

Mathematical Modeling of Competitive Group Recommendation Systems with Application to Peer Review Systems

In this paper, we present a mathematical model to capture various factors which may influence the accuracy of a competitive group recommendation system. We apply this model to peer review systems, i.e., conference or research grants review, which is an essential component in our scientific community. We explore number of important questions, i.e., how will the number of reviews per paper affect the accuracy of the overall recommendation? Will the score aggregation policy influence the final recommendation? How reviewers' preference may affect the accuracy of the final recommendation? To answer these important questions, we formally analyze our model. Through this analysis, we obtain the insight on how to design a randomized algorithm which is both computationally efficient and asymptotically accurate in evaluating the accuracy of a competitive group recommendation system. We obtain number of interesting observations: i.e., for a medium tier conference, three reviews per paper is sufficient for a high accuracy recommendation. For prestigious conferences, one may need at least seven reviews per paper to achieve high accuracy. We also propose a heterogeneous review strategy which requires equal or less reviewing workload, but can improve over a homogeneous review strategy in recommendation accuracy by as much as 30% . We believe our models and methodology are important building blocks to study competitive group recommendation systems.

cs.IR↗

arXiv2011

Online Robust Subspace Tracking from Partial Information

This paper presents GRASTA (Grassmannian Robust Adaptive Subspace Tracking Algorithm), an efficient and robust online algorithm for tracking subspaces from highly incomplete information. The algorithm uses a robust $l^1$-norm cost function in order to estimate and track non-stationary subspaces when the streaming data vectors are corrupted with outliers. We apply GRASTA to the problems of robust matrix completion and real-time separation of background from foreground in video. In this second application, we show that GRASTA performs high-quality separation of moving objects from background at exceptional speeds: In one popular benchmark video example, GRASTA achieves a rate of 57 frames per second, even when run in MATLAB on a personal laptop.

cs.IT↗

arXiv2011

A Matrix-Analytic Solution for Randomized Load Balancing Models with Phase-Type Service Times

In this paper, we provide a matrix-analytic solution for randomized load balancing models (also known as \emph{supermarket models}) with phase-type (PH) service times. Generalizing the service times to the phase-type distribution makes the analysis of the supermarket models more difficult and challenging than that of the exponential service time case which has been extensively discussed in the literature. We first describe the supermarket model as a system of differential vector equations, and provide a doubly exponential solution to the fixed point of the system of differential vector equations. Then we analyze the exponential convergence of the current location of the supermarket model to its fixed point. Finally, we present numerical examples to illustrate our approach and show its effectiveness in analyzing the randomized load balancing schemes with non-exponential service requirements.

cs.NI↗

arXiv2010

Doubly Exponential Solution for Randomized Load Balancing Models with Markovian Arrival Processes and PH Service Times

In this paper, we provide a novel matrix-analytic approach for studying doubly exponential solutions of randomized load balancing models (also known as supermarket models) with Markovian arrival processes (MAPs) and phase-type (PH) service times. We describe the supermarket model as a system of differential vector equations by means of density dependent jump Markov processes, and obtain a closed-form solution with a doubly exponential structure to the fixed point of the system of differential vector equations. Based on this, we show that the fixed point can be decomposed into the product of two factors inflecting arrival information and service information, and further find that the doubly exponential solution to the fixed point is not always unique for more general supermarket models. Furthermore, we analyze the exponential convergence of the current location of the supermarket model to its fixed point, and apply the Kurtz Theorem to study density dependent jump Markov process given in the supermarket model with MAPs and PH service times, which leads to the Lipschitz condition under which the fraction measure of the supermarket model weakly converges the system of differential vector equations. This paper gains a new understanding of how workload probing can help in load balancing jobs with non-Poisson arrivals and non-exponential service times.

cs.DM↗

arXiv2009

On Oligopoly Spectrum Allocation Game in Cognitive Radio Networks with Capacity Constraints

Dynamic spectrum sharing is a promising technology to improve spectrum utilization in the future wireless networks. The flexible spectrum management provides new opportunities for licensed primary user and unlicensed secondary users to reallocate the spectrum resource efficiently. In this paper, we present an oligopoly pricing framework for dynamic spectrum allocation in which the primary users sell excessive spectrum to the secondary users for monetary return. We present two approaches, the strict constraints (type-I) and the QoS penalty (type-II), to model the realistic situation that the primary users have limited capacities. In the oligopoly model with strict constraints, we propose a low-complexity searching method to obtain the Nash Equilibrium and prove its uniqueness. When reduced to a duopoly game, we analytically show the interesting gaps in the leader-follower pricing strategy. In the QoS penalty based oligopoly model, a novel variable transformation method is developed to derive the unique Nash Equilibrium. When the market information is limited, we provide three myopically optimal algorithms "StrictBEST", "StrictBR" and "QoSBEST" that enable price adjustment for duopoly primary users based on the Best Response Function (BRF) and the bounded rationality (BR) principles. Numerical results validate the effectiveness of our analysis and demonstrate the fast convergence of "StrictBEST" as well as "QoSBEST" to the Nash Equilibrium. For the "StrictBR" algorithm, we reveal the chaotic behaviors of dynamic price adaptation in response to the learning rates.

cs.NI↗

arXiv2008

Understanding the Paradoxical Effects of Power Control on the Capacity of Wireless Networks

Recent works show conflicting results: network capacity may increase or decrease with higher transmission power under different scenarios. In this work, we want to understand this paradox. Specifically, we address the following questions: (1)Theoretically, should we increase or decrease transmission power to maximize network capacity? (2) Theoretically, how much network capacity gain can we achieve by power control? (3) Under realistic situations, how do power control, link scheduling and routing interact with each other? Under which scenarios can we expect a large capacity gain by using higher transmission power? To answer these questions, firstly, we prove that the optimal network capacity is a non-decreasing function of transmission power. Secondly, we prove that the optimal network capacity can be increased unlimitedly by higher transmission power in some network configurations. However, when nodes are distributed uniformly, the gain of optimal network capacity by higher transmission power is upper-bounded by a positive constant. Thirdly, we discuss why network capacity in practice may increase or decrease with higher transmission power under different scenarios using carrier sensing and the minimum hop-count routing. Extensive simulations are carried out to verify our analysis.

cs.NI↗