A framework is developed to explore the connection between effective optimization algorithms and the problems they are solving. A number of “no free lunch ” (NFL) theorems are presented which establish that for any algorithm, any elevated performance over one class of problems is offset by performance over another class. These theorems result in a geometric interpretation of what it means for an algorithm to be well suited to an optimization problem. Applications of the NFL theorems to information-theoretic aspects of optimization and benchmark measures of performance are also presented. Other issues addressed include time-varying optimization problems and a priori “head-to-head” minimax distinctions between optimization algorithms, distinctions that result despite the NFL theorems’ enforcing of a type of uniformity over all algorithms.

The canonical particle swarm algorithm is a new approach to optimization, drawing inspiration from group behavior and the establishment of social norms. It is gaining popularity, especially because of the speed of convergence and the fact it is easy to use. However, we feel that each individual is not simply influenced by the best performer among his neighbors. We thus decided to make the individuals “fully informed. ” The results are very promising, as informed individuals seem to find better solutions in all the benchmark functions.

The last few years there has been an increasing amount of interest in the field of distribution estimation optimization algorithms.

As Bayesian networks are applied to more complex and realistic real-world applications, the development of more efficient inference algorithms working under real-time constraints is becoming more and more important. This paper presents a survey of various exact and approximate Bayesian network inference algorithms. In particular, previous research on real-time inference is reviewed. It provides a framework for understanding these algorithms and the relationships between them. Some important issues in real-time Bayesian networks inference are also discussed.

Typically a search problem is posed as an optimisation task in which a cost, or fitness, function must be maximised or minimised, subject to various parameter constraints. This paper discusses "the shape of space", in terms of search algorithms. We point out that it is the combination of representation and traversal operators that define an algorithm's view of a given search problem, and hence gives rise to a fitness landscape. In this sense a fitness landscape only exists in terms of a given search algorithm, and that notions of landscape difficulty are therefore algorithm dependent. We demonstrate these issues for genetic algorithms (GAs) and encoded hill-climbers. We show that both styles of algorithm have attractive features in terms of the way a landscape can be dynamically adjusted, providing a simple means for applying multiple search strategies to a given search problem. We illustrate the techniques for a range of standard cost functions and algorithms. 1. Introduction Recent...

The classic NFL theorems are invariably cast in terms of single objective optimization problems. We confirm that the classic NFL theorem holds for general multiobjective fitness spaces, and show how this follows from a 'single-objective' NFL theorem. We also show that, given any particular Pareto Front, an NFL theorem holds for the set of all multiobjective problems which have that Pareto Front. It follows that, given any 'shape' or class of Pareto fronts, an NFL theorem holds for the set of all multiobjective problems in that class. These findings have salience in test function design. Such NFL results are cast in the typical context of absolute performance, assuming a performance metric which returns a value based on the result produced by a single algorithm. But, in multiobjective search...

There is a developing theory of growing power which, at its current stage of development (indeed, for a number of years now), speaks to qualitative and quantitative aspects of search strategies. Although it has been specialized and applied to genetic algorithms, it's implications and applicability are far more general. This paper deals with the broad outlines of the theory, introducing basic principles and results rather than analyzing or specializing to particular algorithms. A few specific examples are included for illustrative purposes, but the theory's basic structure, as opposed to applications, remains the focus. Key words: Random Heuristic Search, Modeling Evolutionary Algorithms, Degenerate Royal Road Functions. 1 Introduction Vose [20] introduced a rigorous dynamical system model for the binary representation genetic algorithm with proportional selection, mutation determined by a rate, and one-point crossover, using the simplifying assumption of an infinite population. 1 ...

Abstract- The No-Free-Lunch (NFL) theorems hold for general multiobjective fitness spaces, in the sense that, over a space of problems which is closed under permutation, any two algorithms will produce the same set of multiobjective samples. However, there are salient ways in which NFL does not generally hold in multiobjective optimization. Previously we have shown that a ‘free lunch ’ can arise when comparative metrics (rather than absolute metrics) are used for performance measurement. Here we show that NFL does not generally apply in multiobjective optimization when absolute performance metrics are used. This is because multiobjective optimizers usually combine a generator with an archiver. The generator corresponds to the ‘algorithm ’ in the NFL sense, but the archiver filters the sample generated by the algorithm in a way that undermines the NFL assumptions. Essentially, if two multiobjective approaches have different archivers, their average performance may differ. We prove this, and hence show that we can say, without qualification, that some multiobjective approaches are better than others. 1

Leukocytes play an important role in the host defense as they may travel from the blood stream into the tissue in reacting to inflammatory stimuli. The leukocyte-vessel wall interactions are studied in post capillary vessels by intravital video microscopy during in vivo animal experiments. Sequences of video images are obtained and digitized with a frame grabber. A method for automatic detection and characterization of leukocytes in the video images is developed. Individual leukocytes are detected using a neural network that is trained with synthetic leukocyte images generated using a novel stochastic model. This model makes it feasible to generate images of leukocytes with different shapes and sizes under various lighting conditions. Experiments indicate that neural networks trained with the synthetic leukocyte images perform better than networks trained with images of manually detected leukocytes. The best performing neural network trained with synthetic leukocyte images resulted in an 18% larger area under the ROC curve than the best performing neural network trained with manually detected leukocytes.

Informally GA-hardness asks what makes a problem hard or easy for Genetic Algorithms (GAs) to optimize. Characterizing GAhardness has received significant attention since the invention of GAs, yet it remains quite open. In this paper, we first present an abstract, general framework of problem (instance) hardness and algorithm performance for search based on Kolmogorov complexity. We also show, by Rice's theorem, the nonexistence of a predictive GA-hardness measure based only on the description of the problem instance and the configurations of the GA. We then examine several major misconceptions in previous GA-hardness research in the context of this theory. Finally, we propose some promising directions for future research.