Results 1  10
of
14
A tutorial on Bayesian optimization of expensive cost functions, withapplicationtoactiveusermodeling andhierarchical reinforcement learning
, 2009
"... We present a tutorial on Bayesian optimization, a method of finding the maximum of expensive cost functions. Bayesian optimization employs the Bayesian technique of setting a prior over the objective function and combining it with evidence to get a posterior function. This permits a utilitybased se ..."
Abstract

Cited by 29 (2 self)
 Add to MetaCart
We present a tutorial on Bayesian optimization, a method of finding the maximum of expensive cost functions. Bayesian optimization employs the Bayesian technique of setting a prior over the objective function and combining it with evidence to get a posterior function. This permits a utilitybased selection of the next observation to make on the objective function, which must take into account both exploration (sampling from areas of high uncertainty) and exploitation (sampling areas likely to offer improvement over the current best observation). We also present two detailed extensions of Bayesian optimization, with experiments—active user modelling with preferences, and hierarchical reinforcement learning— and a discussion of the pros and cons of Bayesian optimization based on our experiences. 1
A Comparison of Complete Global Optimization Solvers
"... Results are reported of testing a number of existing state of the art solvers for global constrained optimization and constraint satisfaction on a set of over 1000 test problems in up to 1000 variables. ..."
Abstract

Cited by 23 (4 self)
 Add to MetaCart
Results are reported of testing a number of existing state of the art solvers for global constrained optimization and constraint satisfaction on a set of over 1000 test problems in up to 1000 variables.
Adaptive Search with Stochastic Acceptance Probabilities for Global Optimization
"... We present an extension of continuous domain Simulated Annealing. Our algorithm employs a globally reaching candidate generator, adaptive stochastic acceptance probabilities, and converges in probability to the optimal value. An application to simulationoptimization problems with asymptotically dim ..."
Abstract

Cited by 7 (1 self)
 Add to MetaCart
We present an extension of continuous domain Simulated Annealing. Our algorithm employs a globally reaching candidate generator, adaptive stochastic acceptance probabilities, and converges in probability to the optimal value. An application to simulationoptimization problems with asymptotically diminishing errors is presented. Numerical results on a noisy proteinfolding problem are included.
A TimeSensitive System for BlackBox Combinatorial Optimization
 In Workshop on Algorithm Engineering and Experimenation (ALENEX
, 2002
"... When faced with a combinatorial optimization problem, practitioners often turn to blackbox search heuristics such as simulated annealing and genetic algorithms. In blackbox optimization, the problemspecific components are limited to functions that (1) generate candidate solutions, and (2) evaluat ..."
Abstract

Cited by 3 (2 self)
 Add to MetaCart
When faced with a combinatorial optimization problem, practitioners often turn to blackbox search heuristics such as simulated annealing and genetic algorithms. In blackbox optimization, the problemspecific components are limited to functions that (1) generate candidate solutions, and (2) evaluate the quality of a given solution. A primary reason for the popularity of blackbox optimization is its ease of implementation. The basic simulated annealing search algorithm can be implemented in roughly 3050 lines of any modern programming language, not counting the problemspecific localmove and costevaluation functions. This search algorithm is so simple that it is often rewritten from scratch for each new application rather than being reused. In this paper, we examine whether it pays to develop a more sophisticated, generalpurpose heuristic optimization engine. The issue is whether a substantial performance improvement or easeofuse gain results...
The Optimization Test Environment
"... Testing is a crucial part of software development in general, and hence also in mathematical programming. Unfortunately, it is often a time consuming and little exciting activity. This naturally motivated us to increase the e ciency in testing solvers for optimization problems and to automatize as m ..."
Abstract

Cited by 3 (3 self)
 Add to MetaCart
Testing is a crucial part of software development in general, and hence also in mathematical programming. Unfortunately, it is often a time consuming and little exciting activity. This naturally motivated us to increase the e ciency in testing solvers for optimization problems and to automatize as much of the procedure as possible. Keywords: test environment, optimization, solver benchmarking, solver comparison The testing procedure typically consists of three basic tasks: a) organize test problem sets, also called test libraries; b) solve selected test problems with selected solvers; c) analyze, check and compare the results. The Test Environment is a graphical user interface (GUI) that enables to manage the tasks a) and b) interactively, and task c) automatically. The Test Environment is particularly designed for users who seek to 1. adjust solver parameters, or 2. compare solvers on single problems, or 3. evaluate solvers on suitable test sets.
Coloring Graphs With a General Heuristic Search Engine
 Computational Symposium on Graph Coloring and its Generalizations
, 2002
"... this paper, we report on our experiences building implementations of vertex coloring and graph bandwidth coloring with Discropt. We use the same problemindependent heuristic implementations reported in [4] without fundamental changes or problemspeci c tuning. Adapting our heuristics to graph colo ..."
Abstract
 Add to MetaCart
this paper, we report on our experiences building implementations of vertex coloring and graph bandwidth coloring with Discropt. We use the same problemindependent heuristic implementations reported in [4] without fundamental changes or problemspeci c tuning. Adapting our heuristics to graph coloring problems provides a good test of the exibility of our system for two reasons. First, graph coloring is well known as a very challenging problem for local search heuristics such as employed within our system. Second, the fundamental solution representation for graph colorings (set partitions) had not been implemented in our system as of [4]. In this paper, we 1 User Engine Kernel Neighborhood Operator Search Solution Objective Function generates uses uses starts runs Figure 1: Basic system ow test the performance of optimization routines developed for permutation and subset problems on a completely dierent class of object
Quantum Modeling
"... We present a modi cation of Simon's Algorithm 1, 2 that in some cases is able to t experimentally obtained data to appropriately chosen trial functions with high probability. Modulo constants pertaining to the reliability and probability of success of the algorithm, the algorithm runs using only ..."
Abstract
 Add to MetaCart
We present a modi cation of Simon's Algorithm 1, 2 that in some cases is able to t experimentally obtained data to appropriately chosen trial functions with high probability. Modulo constants pertaining to the reliability and probability of success of the algorithm, the algorithm runs using only O(polylog(jY j)) queries to the quantum database and O(polylog(jX j; jY j)) elementary quantum gates where jX j is the size of the experimental data set and jY j is the size of the parameter space. We discuss heuristics for good performance, analyze the performance of the algorithm in the case of linear regression, both onedimensional and multidimensional, and outline the algorithm's limitations.
The Optimization Test Environment User manual
, 2010
"... Abstract. The Test Environment is an interface to efficiently test different optimization solvers. It is designed as a tool for both developers of solver software and practitioners who just look for the best solver for their specific problem class. It enables users to: • Choose and compare diverse s ..."
Abstract
 Add to MetaCart
Abstract. The Test Environment is an interface to efficiently test different optimization solvers. It is designed as a tool for both developers of solver software and practitioners who just look for the best solver for their specific problem class. It enables users to: • Choose and compare diverse solver routines; • Organize and solve large test problem sets; • Select interactively subsets of test problem sets; • Perform a statistical analysis of the results, automatically produced as L ATEX and PDF output. The Test Environment is free to use for research purposes.
Quantum Modeling
, 2003
"... We present a modification of Simon’s Algorithm 1, 2 that in some cases is able to fit experimentally obtained data to appropriately chosen trial functions with high probability. Modulo constants pertaining to the reliability and probability of success of the algorithm, the algorithm runs using only ..."
Abstract
 Add to MetaCart
We present a modification of Simon’s Algorithm 1, 2 that in some cases is able to fit experimentally obtained data to appropriately chosen trial functions with high probability. Modulo constants pertaining to the reliability and probability of success of the algorithm, the algorithm runs using only O(polylog(Y )) queries to the quantum database and O(polylog(X, Y )) elementary quantum gates where X  is the size of the experimental data set and Y  is the size of the parameter space. We discuss heuristics for good performance, analyze the performance of the algorithm in the case of linear regression, both onedimensional and multidimensional, and outline the algorithm’s limitations.