Results 1  10
of
25
Global Optimization of Statistical Functions with Simulated Annealing
 Journal of Econometrics
, 1994
"... Many statistical methods rely on numerical optimization to estimate a model’s parameters. Unfortunately, conventional algorithms sometimes fail. Even when they do converge, there is no assurance that they have found the global, rather than a local, optimum. We test a new optimization algorithm, simu ..."
Abstract

Cited by 281 (2 self)
 Add to MetaCart
Many statistical methods rely on numerical optimization to estimate a model’s parameters. Unfortunately, conventional algorithms sometimes fail. Even when they do converge, there is no assurance that they have found the global, rather than a local, optimum. We test a new optimization algorithm, simulated annealing, on four econometric problems and compare it to three common conventional algorithms. Not only can simulated annealing find the global optimum, it is also less likely to fail on difficult functions because it is a very robust algorithm. The promise of simulated annealing is demonstrated on the four econometric problems.
Minerva: an automated resource provisioning tool for largescale storage systems
 ACM Transactions on Computer Systems
, 2001
"... Enterprisescale storage systems, which can contain hundreds of host computers and storage devices and up to tens of thousands of disks and logical volumes, are difficult to design. The volume of choices that need to be made is massive, and many choices have unforeseen interactions. Storage system d ..."
Abstract

Cited by 131 (24 self)
 Add to MetaCart
(Show Context)
Enterprisescale storage systems, which can contain hundreds of host computers and storage devices and up to tens of thousands of disks and logical volumes, are difficult to design. The volume of choices that need to be made is massive, and many choices have unforeseen interactions. Storage system design is tedious and complicated to do by hand, usually leading to solutions that are grossly overprovisioned, substantially underperforming or, in the worst case, both. To solve the configuration nightmare, we present MINERVA: a suite of tools for designing storage systems automatically. MINERVA uses declarative specifications of application requirements and device capabilities; constraintbased formulations of the various subproblems; and optimization techniques to explore the search space of possible solutions. This paper also explores and evaluates the design decisions that went into MINERVA, using specialized micro and macrobenchmarks. We show that MINERVA can successfully handle a workload with substantial complexity (a decisionsupport database benchmark). MINERVA created a 16disk design in only a few minutes that achieved the same performance as a 30disk system manually designed by human experts. Of equal importance, MINERVA was able to predict the resulting system's performance before it was built.
A hybrid approach for the 0–1 multidimensional knapsack problem
 In Proceedings of the International Joint Conference on Artificial Intelligence 2001
, 2001
"... We present a hybrid approach for the 0–1 multidimensional knapsack problem. The proposed approach combines linear programming and Tabu Search. The resulting algorithm improves significantly on the best known results of a set of more than 150 benchmark instances. 1 ..."
Abstract

Cited by 34 (4 self)
 Add to MetaCart
We present a hybrid approach for the 0–1 multidimensional knapsack problem. The proposed approach combines linear programming and Tabu Search. The resulting algorithm improves significantly on the best known results of a set of more than 150 benchmark instances. 1
Combinatorial Auctions, Knapsack Problems, and Hillclimbing Search
 In Canadian Conference on AI
, 2001
"... . This paper examines the performance of hillclimbing algorithms on standard test problems for combinatorial auctions (CAs). On singleunit CAs, deterministic hillclimbers are found to perform well, and their performance can be improved significantly by randomizing them and restarting them sev ..."
Abstract

Cited by 21 (1 self)
 Add to MetaCart
(Show Context)
. This paper examines the performance of hillclimbing algorithms on standard test problems for combinatorial auctions (CAs). On singleunit CAs, deterministic hillclimbers are found to perform well, and their performance can be improved significantly by randomizing them and restarting them several times, or by using them collectively. For some problems this good performance is shown to be no better than chancel; on others it is due to a wellchosen scoring function. The paper draws attention to the fact that multiunit CAs have been studied widely under a different name: multidimensional knapsack problems (MDKP). On standard test problems for MDKP, one of the deterministic hillclimbers generates solutions that are on average 99% of the best known solutions. 1 Introduction Suppose there are three items for auction, X, Y, and Z, and three bidders, B1, B2, and B3. B1 wants any one of the items and will pay $5, B2 wants two items  X and one of Y or Z  and will pay $9, an...
Cutting and surrogate constraint analysis for improved multidimensional knapsack solutions
 ANNALS OF OPERATIONS RESEARCH
, 2000
"... ... Knapsack Problems to fix some variables to zero and to separate the rest into two groups those that tend to be zero and those that tend to be one, in an optimal integer solution. Using an initial feasible integer solution, we generate logic cuts based on our analysis before solving the problem w ..."
Abstract

Cited by 12 (5 self)
 Add to MetaCart
... Knapsack Problems to fix some variables to zero and to separate the rest into two groups those that tend to be zero and those that tend to be one, in an optimal integer solution. Using an initial feasible integer solution, we generate logic cuts based on our analysis before solving the problem with branch and bound. Computational testing, including the set of problems in the ORlibrary and our own set of difficult problems, shows our approach helps to solve difficult problems in a reasonable amount of time and, in most cases, with a fewer number of nodes in the search tree than leading commercial software. ______________________________________________________________________________________
Oscillation, Heuristic Ordering and Pruning in Neighborhood Search
 In Proceedings of CP'97, G. Smolka ed., LNCS 1330
, 1997
"... . This paper describes a new algorithm for combinatorial optimization problems and presents the results of our experiments. HOLSA Heuristic Oscillating Local Search Algorithm is a neighborhood search algorithm using an evaluation function f inspired from A*, a bestfirst strategy, a pruning of sta ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
(Show Context)
. This paper describes a new algorithm for combinatorial optimization problems and presents the results of our experiments. HOLSA Heuristic Oscillating Local Search Algorithm is a neighborhood search algorithm using an evaluation function f inspired from A*, a bestfirst strategy, a pruning of states as in B&B and operators performing variable steps. All these caracteristics lead to an oscillation principle whereby the search alternates between improving the economic function and satisfying the constraints. We specify how to compute the start state, the evaluation function and the variable steps in order to implement the general outline of HOLSA. Its performance is tested on the multidimensional knapsack problem, using randomly generated problems and classical test problems of the litterature. The experiments show that HOLSA is very efficient, according to the quality of the solutions as well as the search speed, at least on the class of problems studied in this paper. Moreover with ...
Controlling crossover in a selection hyperheuristic framework
, 2011
"... Abstract. In evolutionary algorithms, crossover is used to recombine two candidate solutions to yield a new solution which hopefully inherits good material from both. Hyperheuristics are highlevel search methodologies which operate on a search space of heuristics. Hyperheuristics can be broadly s ..."
Abstract

Cited by 5 (3 self)
 Add to MetaCart
(Show Context)
Abstract. In evolutionary algorithms, crossover is used to recombine two candidate solutions to yield a new solution which hopefully inherits good material from both. Hyperheuristics are highlevel search methodologies which operate on a search space of heuristics. Hyperheuristics can be broadly split into two categories; heuristic selection and generation methodologies. Here we will investigate hyperheuristics from the former category. Selection hyperheuristics select a heuristic to apply from an existing set of lowlevel heuristics at a given point in the search. Crossover is increasingly being included in general purpose hyperheuristic frameworks such as HyFlex and Hyperion however little work has been done to assess how best to utilise it. Since a singlepoint search hyperheuristic operates on a single candidate solution and two candidate solutions are needed for crossover, a mechanism is required to control the choice of the other solution. We propose a framework which maintains a list of potential solutions for use in crossover. We investigate the control of such lists at two levels. Firstly, crossover is controlled at the hyperheuristic level where no problem speci c information is required. Secondly, it is controlled at the problem domain level where problem speci c information is used to produce good quality solutions to use for crossover. A number of selection hyperheuristics are tested over three wellknown benchmark libraries for an NPhard optimisation problem; the multidimensional 01 knapsack problem (MKP). Exact solvers such as CPLEX also use heuristics and have improved signi cantly since the last published application to some of the benchmark data. New results are presented using CPLEX 12.2 over the benchmark instances. *Corresponding author
Genetic Algorithms for 0/1 Multidimensional Knapsack Problems
 Proceedings Norsk Informatikk Konferanse, NIK '96
, 1996
"... An important class of combinatorial optimization problems are the Multidimensional 0/1 Knapsacks, and various heuristic and exact methods have been devised to solve them. Among these, Genetic Algorithms have emerged as a powerful new search paradigms. We show how a proper selection of parameters and ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
An important class of combinatorial optimization problems are the Multidimensional 0/1 Knapsacks, and various heuristic and exact methods have been devised to solve them. Among these, Genetic Algorithms have emerged as a powerful new search paradigms. We show how a proper selection of parameters and search mechanisms lead to an implementation of Genetic Algorithms that yields high quality solutions. The methods are tested on a portfolio of 0/1 multidimensional knapsack problems from literature, and a minimum of domainspecific knowledge is used to guide the search process. The quality of the produced results rivals, and in some cases surpasses, the best solutions obtained by specialpurpose methods that have been created to exploit the special structure of these problems. 1. INTRODUCTION A classic problem from the litterature is the Knapsack Problem, also known as the Burglars Problem. The burglar is given a knapsack which has an upper weight limit of t pounds, and have a choice of it...
A Genetic Programming HyperHeuristic for the Multidimensional Knapsack Problem
"... Abstract—Hyperheuristics are a class of highlevel search techniques which operate on a search space of heuristics rather than directly on a search space of solutions. Early hyperheuristics focussed on selecting and applying a lowlevel heuristic at each stage of a search. Recent trends in hyperhe ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
(Show Context)
Abstract—Hyperheuristics are a class of highlevel search techniques which operate on a search space of heuristics rather than directly on a search space of solutions. Early hyperheuristics focussed on selecting and applying a lowlevel heuristic at each stage of a search. Recent trends in hyperheuristic research have led to a number of approaches being developed to automatically generate new heuristics from a set of heuristic components. This work investigates the suitability of using genetic programming as a hyperheuristic methodology to generate constructive heuristics to solve the multidimensional 01 knapsack problem. A population of heuristics to rank knapsack items are trained on a subset of test problems and then applied to unseen instances. The results over a set of standard benchmarks show that genetic programming can be used to generate constructive heuristics which yield humancompetitive results.