Showing papers on "Incremental heuristic search published in 1996"

Heuristic-biased stochastic sampling

Abstract: This paper presents a search technique for scheduling problems, called Heuristic-Biased Stochastic Sampling (HBSS). The underlying assumption behind the HBSS approach is that strictly adhering to a search heuristic often does not yield the best solution and, therefore, exploration off the heuristic path can prove fruitful. Within the HBSS approach, the balance between heuristic adherence and exploration can be controlled according to the confidence one has in the heuristic. By varying this balance, encoded as a bias function, the HBSS approach encompasses a family of search algorithms of which greedy search and completely random search are extreme members. We present empirical results from an application of HBSS to the real-world problem of observation scheduling. These results show that with the proper bias function, it can be easy to outperform greedy search.

Hierarchical A *: searching abstraction hierarchies efficiently

Abstract: ion, in search, problem solving, and planning, works by replacing one state space by another (the "abstract" space) that is easier to search. The results of the search in the abstract space are used to guide search in the original space. For instance, the length of the abstract solution can be used as a heuristic for A* in searching in the original space. However, there are two obstacles to making this work efficiently. The first is a theorem (Valtorta, 1984) stating that for a large class of abstractions, "embedding abstractions," every state expanded by blind search must also be expanded by A* when its heuristic is computed in this way. The second obstacle arises because in solving a problem A* needs repeatedly to do a full search of the abstract space while computing its heuristic. This paper introduces a new abstraction-induced search technique, "Hierarchical A*," that gets around both of these difficulties: first, by drawing from a different class of abstractions, "homomorphism abstractions," and, secondly, by using novel caching techniques to avoid repeatedly expanding the same states in successive searches in the abstract space. Hierarchical A* outperforms blind search on all the search spaces studied.

Rank ordering and positive bases in pattern search algorithms

Abstract: We present two new classes of pattern search algorithms for unconstrained minimization: the rank ordered and the positive basis pattern search methods. These algorithms can nearly halve the worst case cost of an iteration compared to the classical pattern search algorithms. The rank ordered pattern search methods are based on a heuristic for approximating the direction of steepest descent, while the positive basis pattern search methods are motivated by a generalization of the geometry characteristic of the patterns of the classical methods. We describe the new classes of algorithms and present the attendant global convergence analysis.

Learning equivalence classes of Bayesian network structures

Abstract: Approaches to learning Bayesian networks from data typically combine a scoring function with a heuristic search procedure. Given a Bayesian network structure, many of the scoring functions derived in the literature return a score for the entire equivalence class to which the structure belongs. When using such a scoring function, it is appropriate for the heuristic search algorithm to search over equivalence classes of Bayesian networks as opposed to individual structures. We present the general formulation of a search space for which the states of the search correspond to equivalence classes of structures. Using this space, any one of a number of heuristic search algorithms can easily be applied. We compare greedy search performance in the proposed search space to greedy search performance in a search space for which the states correspond to individual Bayesian network structures.

Local search and the number of solutions

Abstract: There has been considerable research interest into the solubility phase transition, and its effect on search cost for backtracking algorithms. In this paper we show that a similar easy-hard-easy pattern occurs for local search, with search cost peaking at the phase transition. This is despite problems beyond the phase transition having fewer solutions, which intuitively should make the problems harder to solve. We examine the relationship between search cost and number of solutions at different points across the phase transition, for three different local search procedures, across two problem classes (CSP and SAT). Our findings show that there is a significant correlation, which changes as we move through the phase transition.

Using multiple searchers in constrained‐path, moving‐target search problems

Abstract: The search theory open literature has paid little, if any, attention to the multiple-searcher, moving-target search problem. We develop an optimal branch-and-bound procedure and six heuristics for solving constrained-path problems with multiple searchers. Our optimal procedure outperforms existing approaches when used with only a single searcher. For more than one searcher, the time needed to guarantee an optimal solution is prohibitive. Our heuristics represent a wide variety of approaches: One solves partial problems optimally, two use paths based on maximizing the expected number of detections, two are genetic algorithm implementations, and one is local search with random restarts. A heuristic based on the expected number of detections obtains solutions within 2% of the best known for each one-, two-, and three-searcher test problem considered. For one- and two-searcher problems, the same heuristic's solution time is less than that of other heuristics. For three-searcher problems, a genetic algorithm implementation obtains the best-known solution in as little as 20% of other heuristic solution times.

Tabu search for graph partitioning

Abstract: In this paper, we develop a tabu search procedure for solving the uniform graph partitioning problem. Tabu search, an abstract heuristic search method, has been shown to have promise in solving several NP-hard problems, such as job shop and flow shop scheduling, vehicle routing, quadratic assignment, and maximum satisfiability. We compare tabu search to other heuristic procedures for graph partitioning, and demonstrate that tabu search is superior to other solution approaches for the uniform graph partitioning problem both with respect to solution quality and computational requirements.

Genetic Algorithms and Explicit Search Statistics

Abstract: The genetic algorithm (GA) is a heuristic search procedure based on mechanisms abstracted from population genetics. In a previous paper [Baluja & Caruana, 1995], we showed that much simpler algorithms, such as hillclimbing and Population-Based Incremental Learning (PBIL), perform comparably to GAs on an optimization problem custom designed to benefit from the GA's operators. This paper extends these results in two directions. First, in a large-scale empirical comparison of problems that have been reported in GA literature, we show that on many problems, simpler algorithms can perform significantly better than GAs. Second, we describe when crossover is useful, and show how it can be incorporated into PBIL.

New advances in Alpha-Beta searching

Abstract: Alpha-Beta has been the algorithm of choice for game-tree search for over three decades. Its success is largely attributable to a variety of enhancements to the basic algorithm that can dramatically improve the search efficiency. Although state-ofthe-art game-playing programs build trees that are close in size to the minimal Alpha-Beta search tree, this paper shows that there is still room for improvement. Three new enhancements are presented: best-first Alpha-Beta search, better use of transpositions, and improvingaspiration search under real-time constraints. Measurements show that these improvements can reduce search effort by 35%.

Branch-and-Bound Search Algorithms and Their Computational Complexity.

Abstract: : Branch-and-bound (BnB) is a general problem-solving paradigm that has been studied extensively in the areas of computer science and operations research, and has been employed to find optimal solutions to computation-intensive problems. Thanks to its generality, BnB takes many search algorithms, developed for different purposes, as special cases. Some of these algorithms, such as best-first search and depth-first search, are very popular, some, such as iterative deepening, recursive best-first search and constant-space best-first search, are known only in the artificial intelligence area. Because it was studied in different areas, BnB has been described under different formulations. The first part of this paper, we give comprehensive descriptions of the BnB method and of these search algorithms, consolidating the basic features of BnB. In the second part, we summarize recent theoretical development on the average-case complexity of BnB search algorithms.

Using tabu search to solve the Steiner tree-star problem in telecommunications network design

Abstract: We present a study of using tabu search to solve a specific network design problem arising in the telecommunication industry. We develop a tabu search heuristic for this problem that incorporates long-term memory and probabilistic move selections. Computational results show that the new heuristic consistently outperforms the best local search currently available, with significantly increased performance differences on more difficult problems.

Easy and hard testbeds for real-time search algorithms

Abstract: Although researchers have studied which factors influence the behavior of traditional search algorithms, currently not much is known about how domain properties influence the performance of real-time search algorithms. In this paper we demonstrate, both theoretically and experimentally, that Eulerian state spaces (a superset of undirected state spaces) are very easy for some existing real-time search algorithms to solve: even real-time search algorithms that can be intractable, in general, are efficient for Eulerian state spaces. Because traditional real-time search testbeds (such as the eight puzzle and gridworlds) are Eulerian, they cannot be used to distinguish between efficient and inefficient real-time search algorithms. It follows that one has to use non-Eulerian domains to demonstrate the general superiority of a given algorithm. To this end, we present two classes of hard-to-search state spaces and demonstrate the performance of various real-time search algorithms on them.

A hybrid heuristic search algorithm for scheduling flexible manufacturing systems

Abstract: This paper presents a hybrid search algorithm for scheduling flexible manufacturing systems (FMS). The algorithm combines heuristic best-first strategy with controlled backtracking strategy. Timed (place) Petri nets are used for problem representation. Their use allows to explicitly formulate concurrent activities, multiple resources sharing, precedence constraints and dynamic routing in FMS operation. The hybrid heuristic search algorithm is combined with the execution of the timed Petri nets to search for an optimal or near-optimal and deadlock-free schedule. The backtracking strategy is controllable. One can only employ the pure best-first search to obtain an optimal schedule thanks to a proposed admissible heuristic function. The presented method is illustrated through an FMS scheduling problem.

Empirical studies of heuristic local search for constraint solving

Abstract: The goal of this paper is twofold. First, we introduce a class of local search procedures for solving optimization and constraint problems. These procedures are based on various heuristics for choosing variables and values in order to examine a general neighborhood. Second, four combinations of heuristics are empirically evaluated by using the graph-coloring problem and a real world application — the frequency assignment problem. The results are also compared with those obtained with other approaches including simulated annealing, Tabu search, constraint programming and heuristic graph coloring algorithms. Empirical evidence shows the benefits of this class of local search procedures for solving large and hard instances.

Organizational Strategies for Multiagent Real-time Search

Abstract: MultiAgent Real-Time A* (MARTA*) is a multiagent version of Real-Time A* (RTA*) algorithm where multiple agents concurrently and autonomously search and move to find a solution. In this paper, we introduce two organizational strategies based on repulsion and attraction to MArtTA*. Each agent observes the distances from others and moves as it becomes parted from or approaches others, in contrast it simply moves randomly in the original MAR.TA*. Through simulation experiments, we demonstrate repulsion and attraction are effective in both search time and solution quality for Maze and 15-puzzle respectively. However, the opposite combinations of strategy and problem degr’d~le the performance. We finally discuss why the effectiveness of organizational strategy depends on the problem from a viewpoint of heuristic depression. Observed results suggest that there is a fertile research field which crosses over the traditional heuristic search and the organizational approach in multi-agent environment.

Logarithmic convergence of random heuristic search

Abstract: This paper speaks to the inherent emergent behavior of genetic search. For completeness and generality, a class of stochastic search algorithms, random heuristic search, is reviewed. A general convergence theorem for this class is then proved. Since the simple genetic algorithm (GA) is an instance of random heuristic search, a corollary is a result concerning GAs and time to convergence.

Comparing heuristic, evolutionary and local search approaches to scheduling

Abstract: The choice of search algorithm can play a vital role in the success of a scheduling application. In this paper, we investigate the contribution of search algorithms in solving a real-world warehouse scheduling problem. We compare performance of three types of scheduling algorithms: heuristic, genetic algorithms and local search. Additionally, we assess the influence of heuristics on search performance and check for bias induced by using a fast objective function to evaluate intermediate search results.

Navigation within a Structured Search Space in Layout Problems

Abstract: HeGeL-2 is a heuristic-based layout generator equipped with a nonserial dynamic programming algorithm. HeGeL-2 finds optimal solutions to plan-layout design (PLD) problems which are defined as problems that involve the allocation of a nontrivial number (ten or more) of design units in one application. The search strategy HeGeL-2 uses is similar in some respects and dissimilar in others to human search behavior, which has been simulated in an earlier version of HeGeL-2. These similarities include constraint relaxing strategies common to both versions. Dissimilarities concern claims of optimality and ‘goodness’ of solutions measured by an objective function based on soft constraints. In this paper, the general problem of navigation in a search space for layout problems is discussed in the context of a general-purpose formalism. Through this formalism several layout generators, including LOOS, WRIGHT, and HeGeL-2, are described.

Parallel heuristic search - Introductions and a new approach

Abstract: In the previous chapters of this book, we have seen techniques for solving NP-hard combinatorial optimization problems exactly, i.e. methods that nd a solution which is guaranteed to be optimal. These methods carry out an implicit search through the entire space of solutions, thereby guaranteeing the best solution found during this process to be an optimal solution. The price paid for this guarantee is that the running time may increase exponentially with problem size. The unfortunate implication is thus that for many problem types, we are only able to solve small or medium-sized problem instances to proven optimality. A wealth of interesting problems can therefore not be handled by exact methods. A as consequence of this fact, we must in those cases settle for a less ambitious goal: nding solutions which are "good" by some standard, but not necessarily optimal. The techniques for achieving this are either approximation algorithms or heuristics. 1 What are heuristics ? It may be diicult to tell the diierence between approximation algorithms and heuristics, since there is no clear deenition of either of the classes of methods. In this context, we informally deene the bordering line between the two classes as whether or not the algorithm in question provides a performance guarantee with respect to the quality of the produced solution. We thus deene approximation algorithms as algorithms which provide a feasible solution to a problem, with a cost guaranteed to be no more than a given factor higher than the cost of an optimal solution. In 6], Christoodes gives a simple approximation algorithm for the Traveling Salesman Problem (TSP), which produces a tour of length at most a factor 1.5 higher than the cost of an optimal tour. Approximation algorithms are often quite fast, and it is indeed appealing that a guarantee on the solution quality can be given. Still, the bound is seldom particularly tight, as compared to results achieved by heuristics. Contrary to approximation algorithms, heuristics usually only have empirical evidence of their problem solving abilities. Wilf 30] characterizes heuristics as "...methods that seem to work well in practice, for reasons nobody understand...", and this is admittedly true in many cases. Numerous heuristics are tailored to a speciic problem type, one of the most well-known being the Lin-Kernighan

The role of heuristics in automated theorem proving. J.A. Robinson's resolution principle

Abstract: The aim of this paper is to show how J.A. Robinson's resolution principle was perceived and discussed in the AI community between the mid sixties and the first seventies. During this time the so called heuristic search paradigm was still influential in the AI community, and both resolution principle and certain resolution based, apparently human-like, search strategies were matched with those problem solving heuristic procedures which were representative of the AI heuristic search paradigm.

Heuristic reliability optimization by tabu search

Abstract: A new heuristic algorithm, based on the tabu search methodology, is proposed for constrained redundancy optimization in series and in complex systems. It has the advantage of not being blocked as soon as a local optimum is found. Results given by the new method are compared with those of previous heuristics on a series of examples.

Design of fuzzy logic controllers using tabu search algorithm

Abstract: In the design of a fuzzy logic controller for a process an important problem is the construction of the rule base of the controller. This paper describes a new method for constructing the rule base automatically. The method is based on the tabu search algorithm which is a general heuristic procedure for guiding search in a complex search space to find global optimal solutions for difficult problems.

Extreme-Value Search and General Selection Algorithms for Fully Parallel Associative Memories

Abstract: Several useful associative memory algorithms deal with identifying extreme values (maximum or minimum) in a specified field of a selected subset of words. Previously proposed algorithms for such extreme-value searches are bit-sequential in nature, even when implemented on fully parallel associative memories. We show how the multiple-bit search capability of a fully parallel associative memory can be used to advantage in reducing the expected search time for finding extreme values. The idea is to search for the all-ones pattern within subfields of the specified search field in lieu of, or prior to, examining bit slices one at a time. Optimal subfield length is determined for both fixed-size and variable-size bit groupings and the corresponding reduction in search time is quantified. The results are extended to rank-based selection where thejth largest or smallest value in a given field of a selected subset of words is to be identified. We conclude that significant reduction in the number of search cycles is possible in most practical extreme-value search and selection problems.

Simulation optimization through direct search for multi-objective manufacturing systems

Abstract: Simulation modelling has been one of the most widely used techniques for analysing complex manufacturing systems In this paper, we propose a direct search algorithm expanded from the Hooke-Jecves pattern search to systematically and efficiently locate satisfactory solutions for multi-objective simulation models The user-specified goals can be precise and/or fuzzy Heuristic rules stemming from the simulation result of resource statistics are incorporated into the Hooke-Jeeves pattern search The proposed heuristic rules make the search procedure effective regardless of different initial points and various bounded ranges of decision variables Experimental results show that the proposed approach is suitable for analysing complex manufacturing systems, in which multiple objectives and multiple decision variables are encountered