Specific Problem Instances Research Articles

Safe, Strong, and Tractable Relevance Analysis for Planning

In large and complex planning problems, there will almost inevitably be aspects that are not relevant to a specific problem instance. Thus, identifying and removing irrelevant parts from an instance is one of the most important techniques for scaling up automated planning. We examine the path-based relevance analysis method, which is safe (preserves plan existence and cost) and powerful but has exponential time complexity, and show how to make it run in polynomial time with only a minimal loss of pruning power.

Approximate Dynamic Programming via a Smoothed Linear Program

We present a novel linear program for the approximation of the dynamic programming cost-to-go function in high-dimensional stochastic control problems. LP approaches to approximate DP have typically relied on a natural “projection” of a well-studied linear program for exact dynamic programming. Such programs restrict attention to approximations that are lower bounds to the optimal cost-to-go function. Our program—the “smoothed approximate linear program”—is distinct from such approaches and relaxes the restriction to lower bounding approximations in an appropriate fashion while remaining computationally tractable. Doing so appears to have several advantages: First, we demonstrate bounds on the quality of approximation to the optimal cost-to-go function afforded by our approach. These bounds are, in general, no worse than those available for extant LP approaches and for specific problem instances can be shown to be arbitrarily stronger. Second, experiments with our approach on a pair of challenging problems (the game of Tetris and a queueing network control problem) show that the approach outperforms the existing LP approach (which has previously been shown to be competitive with several ADP algorithms) by a substantial margin.

Toward self-describing and workflow integrated Earth system models: A coupled atmosphere-ocean modeling system application

The complexity of Earth system models and their applications is increasing as a consequence of scientific advances, user demand, and the ongoing development of computing platforms, storage systems and distributed high-resolution observation networks. Multi-component Earth system models need to be redesigned to make interactions among model components and other applications external to the modeling system easier. To that end, the common component interfaces of Earth system models can be redesigned to increase interoperability between models and other applications such as various web services, data portals and science gateways. The models can be made self-describing so that the many configuration, build options and inputs of a simulation can be recorded. In this paper, we present a coupled modeling system that includes the proposed methodology to create self-describing models with common model component interfaces. The designed coupled atmosphere-ocean modeling system is also integrated into a scientific workflow system to simplify routine modeling tasks and relationships between these tasks and to demonstrate the enhanced interoperability between different technologies and components. Later on, the work environment is tested using a realistic Earth system modeling application. As can be seen through this example, a layered design for collecting provenance and metadata has the added benefit of documenting a run in far greater detail than before. In this way, it facilitates exploration and understanding of simulations and leads to possible reproducibility. In addition to designing self-describing Earth system models, the regular modeling tasks are also simplified and automated by using a scientific workflow which provides meaningful abstractions for the model, computing environment and provenance/metadata collection mechanisms. Our aim here is to solve a specific instance of a complex model integration problem by using a framework and scientific workflow approach together. The reader may also note that the methods presented in this paper might be also generalized to other types of Earth system models, leading to improved ease of use and flexibility. The initial results also show that the coupled atmosphere-ocean model, which is controlled by the designed workflow environment, is able to reproduce the Mediterranean Sea surface temperature when it is compared with the used CCSM3 initial and boundary conditions.

Impact Assessment of Job Machine Factors on Scaling Parameters

Composite dispatching rules are of great importance in solving realistic scheduling problems because they are fast and easy to implement. The successful implementation of composite dispatching rules depends on their scaling parameter values, while further the appropriate scaling parameter values are determined by specific problem instance features, i.e. job machine factors. In this paper, a Face-centered Cube experimental design is proposed to study the impact of job machine factors on scaling parameter values when Apparent Tardiness Cost with Setups (ATCS) rule is used to schedulePm|sjk|ΣwjTjproblem. The dynamics between the job machine factors and the good scaling parameter values are revealed by experimental results.

Learning Inadmissible Heuristics During Search

Suboptimal search algorithms offer shorter solving times by sacrificing guaranteed solution optimality. While optimal searchalgorithms like A* and IDA* require admissible heuristics, suboptimalsearch algorithms need not constrain their guidance in this way. Previous work has explored using off-line training to transform admissible heuristics into more effective inadmissible ones. In this paper we demonstrate that this transformation can be performed on-line, during search. In addition to not requiring training instances and extensive pre-computation, an on-line approach allows the learned heuristic to be tailored to a specific problem instance. We evaluate our techniques in four different benchmark domains using both greedy best-first search and bounded suboptimal search. We find that heuristics learned on-line result in both faster search andbetter solutions while relying only on information readily available in any best-first search.

Latent Class Models for Algorithm Portfolio Methods

Different solvers for computationally difficult problems such as satisfiability (SAT) perform best on different instances. Algorithm portfolios exploit this phenomenon by predicting solvers' performance on specific problem instances, then shifting computational resources to the solvers that appear best suited. This paper develops a new approach to the problem of making such performance predictions: natural generative models of solver behavior. Two are proposed, both following from an assumption that problem instances cluster into latent classes: a mixture of multinomial distributions, and a mixture of Dirichlet compound multinomial distributions. The latter model extends the former to capture burstiness, the tendency of solver outcomes to recur. These models are integrated into an algorithm portfolio architecture and used to run standard SAT solvers on competition benchmarks. This approach is found competitive with the most prominent existing portfolio, SATzilla, which relies on domain-specific, hand-selected problem features; the latent class models, in contrast, use minimal domain knowledge. Their success suggests that these models can lead to more powerful and more general algorithm portfolio methods.

Privacy-Preserving Outlier Detection Through Random Nonlinear Data Distortion

Consider a scenario in which the data owner has some private or sensitive data and wants a data miner to access them for studying important patterns without revealing the sensitive information. Privacy-preserving data mining aims to solve this problem by randomly transforming the data prior to their release to the data miners. Previous works only considered the case of linear data perturbations--additive, multiplicative, or a combination of both--for studying the usefulness of the perturbed output. In this paper, we discuss nonlinear data distortion using potentially nonlinear random data transformation and show how it can be useful for privacy-preserving anomaly detection from sensitive data sets. We develop bounds on the expected accuracy of the nonlinear distortion and also quantify privacy by using standard definitions. The highlight of this approach is to allow a user to control the amount of privacy by varying the degree of nonlinearity. We show how our general transformation can be used for anomaly detection in practice for two specific problem instances: a linear model and a popular nonlinear model using the sigmoid function. We also analyze the proposed nonlinear transformation in full generality and then show that, for specific cases, it is distance preserving. A main contribution of this paper is the discussion between the invertibility of a transformation and privacy preservation and the application of these techniques to outlier detection. The experiments conducted on real-life data sets demonstrate the effectiveness of the approach.

Agreement Between the Fatty Acid Ethyl Ester Hair Test for Alcohol and Social Workers' Reports

The purpose of this study was to examine the relationship between social worker reports and the fatty acid ethyl ester (FAEE) test as a biomarker for heavy alcohol use. In 2005, a diagnostic program to detect excessive alcohol use by FAEE hair analysis in parents at high risk of having children with fetal alcohol spectrum disorders was established. All cases submitted by Child Protective Services between May and December of 2007 (n = 172) were included comparing social worker reports with FAEE test outcome by odds ratio analysis. A subanalysis of mothers (n = 119), excluding fathers, was also performed. Factors associated with testing positive for hair FAEE in parents, and mothers alone, were: knowledge of a specific instance of problem drinking within the past 6 months (odds ratio [OR] = 5.11, 2.57-10.16 and OR = 8.51, 3.59-20.18, respectively) and third party reports alleging alcohol abuse (OR = 3.31, 1.69-6.46 and OR = 3.30, 1.45-7.50, respectively). Mothers who admitted to heavy drinking were also seven times more likely to test positive for hair FAEE (OR = 6.74, 1.50-30.38) than those who did not. Factors negatively associated with testing positive for hair FAEE in parents, and mothers alone, were: social workers testing for FAEE without the suspicion of alcohol use but rather as a measure to "cover all bases" (OR = 0.09, 0.02-0.40 and (OR = 0.13, 0.03-0.58, respectively) or because of a history/suspicion of illicit drug use (OR = 0.2, 0.07-0.55 and OR = 0.26, 0.08-0.80, respectively). Eleven of 15 reports, indicating levels of consumption, were also in clinical agreement with FAEE test outcome. The FAEE hair test is being applied for the first time in the present context. Our results show the test corroborates well with social workers' suspicion of alcohol use. Reported factors directly related to alcohol use were significantly associated with testing positive for excessive alcohol use, whereas factors not directly related to alcohol use were negatively associated with testing positive.

Robust Controls for Network Revenue Management

Revenue management models traditionally assume that future demand is unknown but can be described by a stochastic process or a probability distribution. Demand is, however, often difficult to characterize, especially in new or nonstationary markets. In this paper, we develop robust formulations for the capacity allocation problem in revenue management using the maximin and the minimax regret criteria under general polyhedral uncertainty sets. Our approach encompasses the following open-loop controls: partitioned booking limits, nested booking limits, displacement-adjusted virtual nesting, and fixed bid prices. In specific problem instances, we show that a booking policy of the type of displacement-adjusted virtual nesting is robust, both from maximin and minimax regret perspectives. Our numerical analysis reveals that the minimax regret controls perform very well on average, despite their worst-case focus, and outperform the traditional controls when demand is correlated or censored. In particular, on real large-scale problem sets, the minimax regret approach outperforms by up to 2% the traditional heuristics. The maximin controls are more conservative but have the merit of being associated with a minimum revenue guarantee. Our models are scalable to solve practical problems because they combine efficient (exact or heuristic) solution methods with very modest data requirements.

Robust Controls for Network Revenue Management

Revenue management models traditionally assume that future demand is unknown but can be described by a stochastic process or a probability distribution. Demand is, however, often difficult to characterize, especially in new or nonstationary markets. In this paper, we develop robust formulations for the capacity allocation problem in revenue management using the maximin and the minimax regret criteria under general polyhedral uncertainty sets. Our approach encompasses the following open-loop controls: partitioned booking limits, nested booking limits, displacement-adjusted virtual nesting, and fixed bid prices. In specific problem instances, we show that a booking policy of the type of displacement-adjusted virtual nesting is robust, both from maximin and minimax regret perspectives. Our numerical analysis reveals that the minimax regret controls perform very well on average, despite their worst-case focus, and outperform the traditional controls when demand is correlated or censored. In particular, on real large-scale problem sets, the minimax regret approach outperforms by up to 2% the traditional heuristics. The maximin controls are more conservative but have the merit of being associated with a minimum revenue guarantee. Our models are scalable to solve practical problems because they combine efficient (exact or heuristic) solution methods with very modest data requirements.

Suppressing the Spread of Email Malcode using Short-term Message Recall

Outbreaks of computer viruses and worms have established a pressing need for developing proactive antivirus solutions. A proactive antivirus solution is one that reliably and accurately detects novel malicious mobile code and one that either prevents damage or recovers systems from the damage that such code inflicts. Research has indicated that behavioral analysis, though provably imprecise, can feasibly predict whether novel behavior poses a threat. Nevertheless, even the most reliable detection methods can conceivably misclassify malicious code or deem it harmful only after substantial damage has taken place. The study of damage control and recovery mechanisms is, therefore, clearly essential to the development of better proactive systems. Earlier work has demonstrated that undoing the damage of malicious code is possible with an appropriate behavior monitoring and recording mechanism. However, it remains that even if a system is recovered, the virulent code may have already propagated to other systems, some of which may not be well-equipped in terms of proactive defenses. Curbing the propagation of undesired code once it has left the boundaries of a system is a hard problem and one that has not received much attention. This work focuses on a specific instance of this difficult problem: viruses and worms that spread by email. In this paper, we explore how advantageous it would be to have a short-term email undo mechanism whose purpose is to recall infected messages. Simulation results demonstrate that such ability can substantially curb the damage of email viruses on a global scale. The results are encouraging because they only assume technology that is either readily available or that is otherwise clearly practical today

Inventory Control with Limited Capacity and Advance Demand Information

Manufacturers make production decisions and carry inventory to satisfy uncertain demand. When holding and shortage costs are high, carrying inventory could be even more expensive for a capacitated production system. Recent developments in information technology and sales strategies enabled firms to acquire, collect, or induce advance demand information. We address a periodic-review, stochastic, capacitated, finite and infinite horizon production system faced by a manufacturer who has the ability to obtain advance demand information. We establish optimal policies and characterize their behavior with respect to capacity, fixed costs, advance demand information, and the planning horizon. With a numerical study, we quantify the value of advance demand information and additional capacity for specific problem instances. We illustrate how advance demand information can be a substitute for capacity and inventory.

Toward reasoning with unified modeling language diagrams based on extensible markup language declarative description theory

A practical framework for representing knowledge and reasoning in the domain of Unified Modeling Language (UML) is proposed. In this framework, graphical diagrams in a UML model are encoded as Extensible Markup Language (XML)/Metadata Interchange (XMI) elements, which are regarded as facts about a specific problem instance in a knowledge base, and the general knowledge on UML, such as inherent interrelationships among diagram components and implicit properties of diagrams, is represented as a set of XML definite clauses. Equivalent transformation (ET) is used as a fundamental computation mechanism for reasoning with the represented diagrams. © 2004 Wiley Periodicals, Inc.

Two-level branch prediction using neural networks

Dynamic branch prediction in high-performance processors is a specific instance of a general time series prediction problem that occurs in many areas of science. Most branch prediction research focuses on two-level adaptive branch prediction techniques, a very specific solution to the branch prediction problem. An alternative approach is to look to other application areas and fields for novel solutions to the problem. In this paper, we examine the application of neural networks to dynamic branch prediction. We retain the first level history register of conventional two-level predictors and replace the second level PHT with a neural network. Two neural networks are considered: a learning vector quantisation network and a backpropagation network. We demonstrate that a neural predictor can achieve misprediction rates comparable to conventional two-level adaptive predictors and suggest that neural predictors merit further investigation.

Constraint programming approach for school timetabling

In this paper, the timetabling problem for a typical high school environment was modeled and solved using a constraint programming (CP) approach. In addition, operations research (OR) models and local search techniques were also used in order to assist the CP search process by effectively reducing the solution search space. Relaxed models that can be solved using minimum cost matching algorithms were used in order to calculate problem lower bounds at various instances of the solution process. These bounds were in turn used to prioritize the search options of the CP process. The use of minimum cost matching model in the search process is an economical and efficient mechanism for the creation of effective search strategies and it is a competitive manner of introducing problem domain information in the CP environment. By including in the solution process a sequence of local search steps, the solution quality was further improved. Several large problems were solved and actual computational results for specific problem instances are presented. Scope and purpose There exist various school timetabling problems depending on the environment and the characteristics of the particular school level [1]. In this paper the high school situation in which the teachers teach in several different class sections during the day and the students remain in their classrooms is modeled and solved. The objective function attempts to minimize the idle hours between the daily teaching responsibilities of all the teachers while also attempting to satisfy their requests for early or late shift assignments. The school timetabling problem is combinatorial and there are several strict organizational and sequence-related rules that must be respected. The problem specifications used in this paper, although they closely describe the situation of a typical Greek high school, are quite general and abstract, which makes the findings of this paper applicable to wider set of school timetabling problems. The specifications mainly focus on the fact that each teacher is scheduled to lecture for a given number of hours at a fixed subset of class sections and the requirement that all the class sections must be always in session without any empty periods in their daily schedules. The integration of constraint programming and operations research techniques for the solution of this problem is one of the main contributions of this paper. The solutions obtained fully utilized the data management and organizational capabilities of the constraint programming approach while being assisted in the search path selection process by techniques and algorithms form the operations research pool of knowledge. The additional information provided by the calculation of efficient lower bounds and the subproblem domain definition and solution strategy presented in this paper, further assists the CP process in selecting promising search paths.

Balancing Different Interests in Aesthetic Controls

Aesthetic controls present a specific instance of a general problem faced by planners: balancing the interests of an individual (in altering the physical environments they in-habit) against the interests of the collective (in protecting the visual quality of cities). In this article, the authors discuss the use of aesthetic controls in the United States and review the case for and against restricting architectural expression. The authors then examine two questions: what is the rationale for protecting such expression, and to what extent is this protection warranted? They find two very strong but conflicting interests—one in controlling design expression and the other in not doing so. How might this conflict be resolved? The authors propose a middle ground that is characterized by four criteria: a clearly articulated and demonstrable public interest, demonstrable links to the stated intent, application early in the design or decision process, and allowing and encouraging a variety of acceptable decisions.

Constraint propagation techniques for the disjunctive scheduling problem

Constraint propagation is an elementary method for reducing the search space of combinatorial search and optimization problems which has become more and more important in the last decades. The basic idea of constraint propagation is to detect and remove inconsistent variable assignments that cannot participate in any feasible solution through the repeated analysis and evaluation of the variables, domains and constraints describing a specific problem instance. The contribution of this paper is twofold. The first contribution is a description of efficient constraint propagation methods also known as consistency tests for the disjunctive scheduling problem (DSP) which is a generalization of the classical job shop scheduling problem (JSP). Applying an elementary constraint based approach involving a limited number of search variables, we will derive consistency tests that ensure 3 -b-consistency. We will further present and analyze both new and classical consistency tests which to some extent are generalizations of the aforementioned consistency tests involving a higher number of variables, but still can be implemented efficiently with a polynomial time complexity. Further, the concepts of energetic reasoning and shaving are analyzed and discussed. The other contribution is a classification of the consistency tests derived according to the domain reduction achieved. The particular strength of using consistency tests is based on their repeated application, so that the knowledge derived is propagated, i.e., reused for acquiring additional knowledge. The deduction of this knowledge can be described as the computation of a fixed point. Since this fixed point depends upon the order of the application of the tests, we first derive a necessary condition for its uniqueness. We then develop a concept of dominance which enables the comparison of different consistency tests as well as a simple method for proving dominance. An extensive comparison of all consistency tests is given. Quite surprisingly, we will find out that some apparently stronger consistency tests are subsumed by apparently weaker ones. At the same time an open question regarding the effectiveness of energetic reasoning is answered.

Adaptive Scheduling of Computations and Communications on Distributed-Memory Systems

Compile-time scheduling is one approach to extract parallelism which has proved effective when the execution behavior is predictable. Unfortunately, the performance of most priority-based scheduling algorithms is computation dependent. Scheduling based on the concept of earliest-startable-task produces reasonably short schedules only when available parallelism is large enough to cover the communications. A priority-based decision is more effective when parallelism is low. We propose a scheduling in which the decision function combines two concepts: (1) task-level as global priority and (2) earliest-task-first as local priority The degree of dominance of one of the above concepts is controlled by a computation profile factor that is the ratio of task parallelism to communication. It is shown that the above factor is an upper bound on the deviation of schedule length from optimum. To tune the solution finish time the above scheduler is iteratively applied on the computation graph. In each iteration, the newly generated schedule is used to sharpen the task-levels which contribute in finding shorter schedules in the next iteration. Evaluation is carried out for a wide category of computation graphs with communications for which optimum schedules are known. It is found that pure local scheduling and static priority-based scheduling significantly deviate from the optimum under specific problem instances. Our approach to adapting the scheduling decision to the computation profile is able to produce near-optimum solutions via a much reduced number of iterations than other approaches.

The aim of psychoanalytic therapy.

THE AIM OF PSYCHOANALYTIC THERAPY* 1 Within the broad expanse of the discipline of medicine, which aims at the preservation and restoration of health and the prolongation of life, there exists the subdivision of psychiatry whose special field of endeavor is the furtherance of mental health. Because the study of mental health embraces all that is human in the widest meaning of the word, it has remained difficult for the psychiatrist to state his objectives, describe his techniques and define his subject matter with the precision and specificity which generally prevail in other divisions of the biological sciences. It has been a peculiarity of psychotherapy in particular, which considers everything pertaining to mankind as legitimate grist for its mill, that a statement of its aims and of an economical procedure for their achievement has not been set forth with that clarity which commends itelf to the scientific mind. It is certainly true that, with the continuing refinement of psychoanalytic technique and the ever accumulating insight into human nature which has resulted therefrom, psychotherapy has achieved a direction and purposiveness which justify its claim as a logical method for the treatment of personality disorders. Nevertheless, the psychoanalyst finds himself rather frequently beset by a disconcerting difficulty in the accurate delineation of the aims toward which his efforts are directed, although this difficulty is only a specific instance of the general problem to which mankind will probably always be heir, namely, the question of how to lead the good life. If the goals toward which the psychoanalyst strives are not always as clear in his own mind as he would like them to be, how much more so must this be the case of the unhappy victim of psychoneurosis or other emotional disorder who seeks his aid? Indeed, the patient who presents himself for psychoanalytic treatment and who invariably has only a skimpy familiarity with the details of his illness, feels only too clearly in his own mind the chief end result of his disability, namely, that he is drifting through life in a rudderless manner, not knowing what he wants nor how to proceed in an effective manner toward the few evanescent goals which occasionally appear attractive, however temporarily, to him. While it is plain that, in greater or less degree, the psychoanalyst has been successful in finding practical methods of presenting both to himself and to his patient the aims of the psychotherapeutic endeavor, it seems highly desirable to have a simplified statement of goal to which both patient and physician can subscribe, comparable to that of general medicine, viz., health. Such a simple statement of the hallmark of the well ordered mental economy, which is acceptable to both physician and patient, which is in fact constantly striven for in all psychotherapeutic endeavors, is the achievement of happiness. While both patient and psychoanalyst will probably be quick to agree that the ultimate goal of their cooperative efforts is the achievement of durable happiness for the patient insofar as reality permits, there is likely to be considerable difference of opinion as to what constitutes happiness, which forces operate in favor of it and which militate against it, and what is the most effective method for the attainment of that desired state. The answers to these questions will vary according to the discipline which is being brought to bear upon the problem; but any discipline which purports to institute or enhance a state of happiness will contribute something to the desired aim, provided its tenets are based on trustworthy evidence, and, in theory at least, these varied contributions should be susceptible to the employ of the psychoanalyst. Thus the cleric would probably state that happiness is to be gained through love, that the greatest forces available during the quest for the loving state are faith in the fact that love exists and hope that one will find it, while the most effective technique for gaining happiness is the practice of charity, which is repetitive exercise in the act of loving. …

A hybrid GA-SA algorithm for just-in-time scheduling of multi-level assemblies

The environment is a manufacturing facility that produces multi-level assemblies in a Just-In-Time (JIT) fashion. The due-dates and lot-sizes of the end-items are given, and the objective is to determine a lot-for-lot operations schedule that minimizes the cumulative production lead-time. The scheduling problem within such an environment is NP-hard, and therefore, the performance of heuristics may vary depending on the specific problem instance. To address this problem an effective hybrid Genetic Algorithm-Simulated Annealing (GA-SA) algorithm is developed. The GA starts with an initial population generated by well known scheduling heuristics, a critical path heuristic, and randomly generated schedules. The scheduling work is shared by the GA and SA in two phases that alternate until convergence: (1) Phase I is the GA that crosses over solutions for different work-centers, and (2) Phase II is the SA that improves the sequence of operations on individual work-centers. The effectiveness of the proposed heuristic is assessed via numerical studies.