Parallel noising methods embedded in an adaptive memory

2003

Parallel implementations of metaheuristics appear quite naturally as an effective alternative to speed up the search for approximate solutions of combinatorial optimization problems. They not only allow solving larger problems or finding improved solutions with respect to their sequential counterparts, but they also lead to more robust algorithms. We review some trends in parallel computing and report recent results about linear speedups that can be obtained with parallel implementations using multiple independent processors. Parallel implementations of tabu search, GRASP, genetic algorithms, simulated annealing, and ant colonies are reviewed and discussed to illustrate the main strategies used in the parallelization of different metaheuristics and their hybrids. * 2. Current trends in parallel computing. In this section, we review some trends and fundamental issues in parallel computing, which interfere in the design and in the efficiency of parallel algorithms.

Sixth International Conference on Parallel and Distributed Computing Applications and Technologies (PDCAT'05), 2005

This paper describes how to solve very large Traveling- Salesman Problems heuristically by the parallelization of self-organizing maps on cluster architectures. The used way of parallelizing is a sophisticated Structural Data Parallel approach based on the SPMD model. We distinguish between a non-sophisticated and a sophisticated approach for efficient and simple parallelization of the SOMs.

… of 4th Meta-heuristics International Conference, 2001

In recent years, there have been significant advances in the theory and application of metaheuristics to approximate solutions of complex optimization problems. The $\mathrm{m}\mathrm{e}\mathrm{t}\mathrm{a}-\mathrm{h}\mathrm{e}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{s}\mathrm{t}\dot{\mathrm{i}}\mathrm{c}\mathrm{s}$ term was used: as a language and a program for stating and solving combinatorial problems in [1]; to describe tabu search in [2] and [3]; to classify recent approaches such as adaptive memory programming, ants systems, evolutionary methods, genetics algorithm, greedy randomized adaptive search procedures, guided local search, neural networks, problem-space search, simulated annealing, scatter search, tabu search, threshold algorithms, and their hybrids in [4, 5, 6 and 7] and as a title for the biennial series of the metaheuristics international conferences (MIC-95, MIC-97, MIC-99, MIC-01). A metaheuristic was defined in $[7, 8]$ an iterative master process that guides and modifies the operations of subordinate heuristics to efficiently produce high quality solutions. It may combine intelligently different concepts for exploring the search space and uses learning strategies to structure information. It may manipulate a complete (or incomplete) single solution or a collections of solutions at each iteration. The subordinate heuristics may be high ($\mathrm{o}\mathrm{I}^{\cdot}$ low) level procedures, or a simple local search, or just a construction method. Metaheuristics provide decision makers with robust tools that obtain high quality solutions, in a reasonable computational effort, to important applications in business, engineering, economics and the sciences. Finding exact solutions to these applications still poses a real challenge despite the impact of recent advances in computer technology and the great interaction between computer science, management $\mathrm{s}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}/\mathrm{o}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{s}$ research and mathematics. For more details on theory and applications, we refer to the comprehensive bibliography on metaheuristics in [5], the books in [6-16]. A metaheuristic may have four components: initial space of solutions; search engines; learning and guideline strategies; management of information structures. In this paper, the most efficient metaheuristics and their associated components are briefly described. The unified-metaheuristic framework presented [4] is extended into a more general one to show how the existing metaheuristics can fit into it. The general framework invites extra research into desiging new innovative and unexplored metaheuristics. Finally, we conclude by highlighting current trends and future research directions in this active area of the science of heuristics.

Ant colony optimization is one of the most successful examples of swarm intelligent systems. The exploration and exploitation are the main mechanisms in controlling search within the ACO. Reactive search is a framework for automating the exploration and exploitation in stochastic algorithms. Restarting the search with the aid of memorizing the search history is the soul of reaction. It is to increase the exploration only when needed. This paper proposes a reactive memory model to overcome the limitation of the random exploration after restart because of losing the previous history of search. The proposed model is utilized to record the previous search regions to be used as reference for ants after restart. The performances of six (6) ant colony optimization variants were evaluated to select the base for the proposed model. Based on the results, Max-Min Ant System has been chosen as the base for the modification. The modified algorithm called RMMAS, was applied to TSPLIB95 data and results showed that RMMAS outperformed the standard MMAS.

International Journal of Production Economics - INT J PROD ECON, 1991

DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal. If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the "Taverne" license above, please follow below link for the End User Agreement:

Intelligent Systems Design and …, 2009

In this paper we consider the problem of finding a global optimum of an unconstrained multimodal function within the framework of adaptive memory programming, focusing on an integration of the Scatter Search and Search methodologies. Computational comparisons are performed on a test-bed of 11 types of problems. For each type, four problems are considered, each one with dimension 50, 100, 200 and 500 respectively; thus totalling 44 instances. Our results show that the Scatter Tabu Search procedure is competitive with the state-of-the-art methods in terms of the average optimality gap achieved.

Lecture Notes in Computer Science, 2000

A memory-based heuristic is a function, h(s), stored in the form of a lookup table: h(s) is computed by mapping s to an index and then retrieving the corresponding entry in the table. In this paper we present a notation for describing state spaces, PSVN, and a method for automatically creating memory-based heuristics for a state space by a bstracting its PSVN description. Two i n vestigations of these automatically generated heuristics are presented. First, thousands of automatically generated heuristics are used to experimentally investigate the conjecture by Korf 4] that m t is a constant, where m is the size of a heuristic's lookup table and t is the number of nodes expanded when the heuristic is used to guide search. Second, a similar large-scale experiment i s u s e d t o v erify that the Korf and Reid's complexity analysis 5] can be used to rapidly and reliably choose the best among a given set of heuristics.

2005

We have explored in this paper the parallel implementation of Kohonen's Self Organizing Map (SOM) in simultaneous multithreading architectures. A new method is proposed that outperforms classic map-partitioning approaches targeted for shared-memory multiprocessors. As an application study we have chosen an image classification problem in 3D Electron Microscopy. Performance results are taken using real biological data on an hyperthreading-enabled Intel Pentium 4.

In this article we briefly review the central constructs in combinatorial optimization and in statistical mechanics and then develop the similarities between the two fields. We show how the Metropolis algorithm for approximate numerical simulation of the behavior of a manybody system at a finite temperature provides a natural tool for bringing the techniques of statistical mechanics to bear on optimization.