Results 1  10
of
105
Optimization Flow Control, I: Basic Algorithm and Convergence
 IEEE/ACM TRANSACTIONS ON NETWORKING
, 1999
"... We propose an optimization approach to flow control where the objective is to maximize the aggregate source utility over their transmission rates. We view network links and sources as processors of a distributed computation system to solve the dual problem using gradient projection algorithm. In thi ..."
Abstract

Cited by 518 (55 self)
 Add to MetaCart
We propose an optimization approach to flow control where the objective is to maximize the aggregate source utility over their transmission rates. We view network links and sources as processors of a distributed computation system to solve the dual problem using gradient projection algorithm. In this system sources select transmission rates that maximize their own benefits, utility minus bandwidth cost, and network links adjust bandwidth prices to coordinate the sources' decisions. We allow feedback delays to be different, substantial and timevarying, and links and sources to update at different times and with different frequencies. We provide asynchronous distributed algorithms and prove their convergence in a static environment. We present measurements obtained from a preliminary prototype to illustrate the convergence of the algorithm in a slowly timevarying environment.
Prioritized sweeping: Reinforcement learning with less data and less time
 Machine Learning
, 1993
"... We present a new algorithm, Prioritized Sweeping, for e cient prediction and control of stochastic Markov systems. Incremental learning methods such asTemporal Di erencing and Qlearning have fast real time performance. Classical methods are slower, but more accurate, because they make full use of ..."
Abstract

Cited by 316 (5 self)
 Add to MetaCart
We present a new algorithm, Prioritized Sweeping, for e cient prediction and control of stochastic Markov systems. Incremental learning methods such asTemporal Di erencing and Qlearning have fast real time performance. Classical methods are slower, but more accurate, because they make full use of the observations. Prioritized Sweeping aims for the best of both worlds. It uses all previous experiences both to prioritize important dynamic programming sweeps and to guide the exploration of statespace. We compare Prioritized Sweeping with other reinforcement learning schemes for a number of di erent stochastic optimal control problems. It successfully solves large statespace real time problems with which other methods have di culty. 1 1
A MarketOriented Programming Environment and its Application to Distributed Multicommodity Flow Problems
 Journal of Artificial Intelligence Research
, 1993
"... Market price systems constitute a wellunderstood class of mechanisms that under certain conditions provide effective decentralization of decision making with minimal communication overhead. In a marketoriented programming approach to distributed problem solving, we derive the activities and resour ..."
Abstract

Cited by 274 (21 self)
 Add to MetaCart
Market price systems constitute a wellunderstood class of mechanisms that under certain conditions provide effective decentralization of decision making with minimal communication overhead. In a marketoriented programming approach to distributed problem solving, we derive the activities and resource allocations for a set of computational agents by computing the competitive equilibrium of an artificial economy. Walras provides basic constructs for defining computational market structures, and protocols for deriving their corresponding price equilibria. In a particular realization of this approach for a form of multicommodity flow problem, we see that careful construction of the decision process according to economic principles can lead to efficient distributed resource allocation, and that the behavior of the system can be meaningfully analyzed in economic terms. 1. Distributed Planning and Economics In a distributed or multiagent planning system, the plan for the system as a whole i...
LogGP: Incorporating Long Messages into the LogP Model  One step closer towards a realistic model for parallel computation
, 1995
"... We present a new model of parallel computationthe LogGP modeland use it to analyze a number of algorithms, most notably, the single node scatter (onetoall personalized broadcast). The LogGP model is an extension of the LogP model for parallel computation [CKP + 93] which abstracts the comm ..."
Abstract

Cited by 236 (1 self)
 Add to MetaCart
We present a new model of parallel computationthe LogGP modeland use it to analyze a number of algorithms, most notably, the single node scatter (onetoall personalized broadcast). The LogGP model is an extension of the LogP model for parallel computation [CKP + 93] which abstracts the communication of fixedsized short messages through the use of four parameters: the communication latency (L), overhead (o), bandwidth (g), and the number of processors (P ). As evidenced by experimental data, the LogP model can accurately predict communication performance when only short messages are sent (as on the CM5) [CKP + 93, CDMS94]. However, many existing parallel machines have special support for long messages and achieve a much higher bandwidth for long messages compared to short messages (e.g., IBM SP2, Paragon, Meiko CS2, Ncube/2). We extend the basic LogP model with a linear model for long messages. This combination, which we call the LogGP model of parallel computation, has o...
On the DouglasRachford splitting method and the proximal point algorithm for maximal monotone operators
, 1992
"... ..."
Locally Weighted Learning for Control
, 1996
"... Lazy learning methods provide useful representations and training algorithms for learning about complex phenomena during autonomous adaptive control of complex systems. This paper surveys ways in which locally weighted learning, a type of lazy learning, has been applied by us to control tasks. We ex ..."
Abstract

Cited by 159 (17 self)
 Add to MetaCart
Lazy learning methods provide useful representations and training algorithms for learning about complex phenomena during autonomous adaptive control of complex systems. This paper surveys ways in which locally weighted learning, a type of lazy learning, has been applied by us to control tasks. We explain various forms that control tasks can take, and how this affects the choice of learning paradigm. The discussion section explores the interesting impact that explicitly remembering all previous experiences has on the problem of learning to control.
UtilityBased Rate Control in the Internet for Elastic Traffic
, 2002
"... In a communication network, a good rate allocation algorithm should reflect the utilities of the users while being fair. We investigate this fundamental problem of achieving the system optimal rates in the sense of maximizing aggregate utility, in a distributed manner, using only the information ava ..."
Abstract

Cited by 68 (3 self)
 Add to MetaCart
In a communication network, a good rate allocation algorithm should reflect the utilities of the users while being fair. We investigate this fundamental problem of achieving the system optimal rates in the sense of maximizing aggregate utility, in a distributed manner, using only the information available at the end hosts of the network. This is done by decomposing the overall system problem into subproblems for the network and for the individual users by introducing a pricing scheme. The users are to solve the problem of maximizing individual net utility, which is the utility less the amount they pay. We provide algorithms for the network to adjust its prices and the users to adjust their window sizes such that at an equilibrium the system optimum is achieved. Further, the equilibrium prices are such that the system optimum achieves weighted proportional fairness. It is notable that the update algorithms of the users do not require any explicit feedback from the network, rendering them easily deployable over the Internet. Our scheme is incentive compatible in that there is no benefit to the users to lie about their utilities.
Understanding TCP Vegas: A Duality Model
 In Proceedings of ACM Sigmetrics
, 2001
"... This paper presents a model of the TCP Vegas congestion control mechanism as a distributed optimization algorithm. Doing so has three important benefits. First, it helps us gain a fundamental understanding of why TCP Vegas works, and an appreciation of its limitations. Second, it allows us to prove ..."
Abstract

Cited by 66 (2 self)
 Add to MetaCart
This paper presents a model of the TCP Vegas congestion control mechanism as a distributed optimization algorithm. Doing so has three important benefits. First, it helps us gain a fundamental understanding of why TCP Vegas works, and an appreciation of its limitations. Second, it allows us to prove that Vegas stabilizes at a weighted proportionally fair allocation of network capacity when there is sufficient buffering in the network. Third, it suggests how we might use explicit feedback to allow each Vegas source to determine the optimal sending rate when there is insufficient buffering in the network. We present simulation results that validate our conclusions.
On Asynchronous Iterations
, 2000
"... Asynchronous iterations arise naturally parallel computers wants minimize times. This paper reviews certain models asynchronous iterations, using a common theoretical framework. The corresponding convergence theory and various domains applications presented. These include nonsingular linear systems, ..."
Abstract

Cited by 57 (11 self)
 Add to MetaCart
Asynchronous iterations arise naturally parallel computers wants minimize times. This paper reviews certain models asynchronous iterations, using a common theoretical framework. The corresponding convergence theory and various domains applications presented. These include nonsingular linear systems, nonlinear systems, initial value problems.