Timo Mantere

Solving, rating and generating Sudoku puzzles with GA

This paper studies the problems involved in solving, rating and generating Sudoku puzzles with genetic algorithms (GA). Sudoku is a number puzzle that has recently become a worldwide phenomenon. Sudoku can be regarded as a constraint satisfaction problem. When solved with genetic algorithms it can be handled as a multi-objective optimization problem. The three objectives of this study was: 1. to test if genetic algorithm optimization is an efficient method for solving Sudoku puzzles, 2. can GA be used to generate new puzzles efficiently, and 3. can GA be used as a rating machine that evaluates the difficulty of a given Sudoku puzzle. The last of these objectives is approached by testing if puzzles that are considered difficult for a human solver are also difficult for the genetic algorithm. The results presented in this paper seem to support the conclusion that these objectives are reasonably well met with genetic algorithm optimization.

Genetic Algorithm Based Software Testing

In this work we axe studying possibilities to test software using genetic algorithm search. The idea is to produce test cases in order to find problematic situations like processing time extremes. The proposed test method comes under the heading of automated dynamic stress testing.

A Review of Optical Nondestructive Visual and Near-Infrared Methods for Food Quality and Safety

This paper is a review of optical methods for online nondestructive food quality monitoring. The key spectral areas are the visual and near-infrared wavelengths. We have collected the information of over 260 papers published mainly during the last 20 years. Many of them use an analysis method called chemometrics which is shortly described in the paper. The main goal of this paper is to provide a general view of work done according to different FAO food classes. Hopefully using optical VIS/NIR spectroscopy gives an idea of how to better meet market and consumer needs for high-quality food stuff.

Solving and analyzing Sudokus with cultural algorithms

This paper studies how cultural algorithm suits to solving and analyzing Sudoku puzzles. Sudoku is a number puzzle that has recently become a worldwide phenomenon. It can be regarded as a combinatorial problem, but when solved with evolutionary algorithms it can also be handled as a constraint satisfaction or multi-objective optimization problem. The objectives of this study were 1) to test if a cultural algorithm with a belief space solves Sudoku puzzles more efficiently than a normal permutation genetic algorithm, 2) to see if the belief space gathers information that helps analyze the results and improve the method accordingly, 3) to improve our previous Sudoku solver presented in CEC2007. Experiments showed that proposed the cultural algorithm performed slightly better than the previous genetic algorithm based Sudoku solver.

Improved ant colony genetic algorithm hybrid for Sudoku solving

In this paper we introduce our new improved version of ant colony optimization/genetic algorithm hybrid for Sudoku puzzle solving. Sudoku is combinatorial number puzzle that had become worldwide phenomenon in the last decade. It has also become popular mathematical test problem in order to test new optimization ideas and algorithms for combinatorial problems. In this paper we present our new ideas for populations sorting and elitism rules in order to improve our earlier evolutionary algorithm based Sudoku solvers. Experimental results show that the new ideas significantly improved the speed of Sudoku solving.

Searching protection relay response time extremes using genetic algorithm—software quality by optimization

In this work we study possibilities to automate the searching and measuring of response time extremes of protection relay software using genetic algorithm optimization. The idea is to produce test cases to reveal potentially problematic situations causing processing time extremes in the software of an electric network protection relay. The testing was done using a relay software in a simulator environment. In the comparison performed, the genetic algorithm-based method was clearly better than a pure random test method.

Solving Optical Skin Simulation Model Parameters Using Genetic Algorithm

Near infrared spectroscopy is noninvasive method to obtain information from materials, such as human skin. A simulation model of light interaction with skin is used to simulate skin reflectance spectra when the chemical and physical parameters of the skin are known. Genetic algorithm is utilised for tuning the simulator to solve the inverse problem; to calculate the skin parameters from the measured reflectance spectra. The inverse problems are often ill-posed, which was also true for this problem in its original form. After assuming all physical parameters as fixed, the problem was regularised and a unique solution for blood melanin and water concentrations was found in all simulations. The accuracy and the uniqueness of the solution proved to be almost independent of the provided spectral resolution, as long as it is larger than three wavelengths. The accuracy of the solution depends on the MCML simulation noise level and the fitness function used. The performance of four different fitness functions was evaluated using fitness landscape and noise analysis, and the best of them was chosen. The achieved accuracy is satisfactory for many applications and it can probably be further improved by increasing the number of photons used in the MCML simulation or by further optimising the fitness function. Index Terms—Genetic algorithms, Monte Carlo methods, In- verse problems, Biomedical engineering, Optics, Biological tis- sues.

Searching strain field parameters by genetic algorithms

This paper studies the applicability of genetic algorithms and imaging to measure deformations. Genetic algorithms are used to search for the strain field parameters of images from a uniaxial tensile test. The non-deformed image is artificially deformed according to the estimated strain field parameters, and the resulting image is compared with the true deformed image. The mean difference of intensities is used as a fitness function. Results are compared with a node-based strain measurement algorithm developed by Koljonen et al. The reference method slightly outperforms the genetic algorithm as for mean difference of intensities. The root-mean-square difference of the displacement fields is less than one pixel. However, with some improvements suggested in this paper the genetic algorithm based method may be worth considering, also in other similar applications: Surface matching instead of individual landmarks can be used in camera calibration and image registration. Search of deformation parameters by genetic algorithms could be applied in pattern recognition tasks e.g. in robotics, object tracking and remote sensing if the objects are subject to deformation. In addition, other transformation parameters could be simultaneously looked for.

Threshold matrix for digital halftoning by genetic algorithm optimization

Digital halftoning is used both in low and high resolution high quality printing technologies. Our method is designed to be mainly used for low resolution ink jet marking machines to produce both gray tone and color images. The main problem with digital halftoning is pink noise caused by the human eyes visual transfer function. To compensate for this the random dot patterns used are optimized to contain more blue than pink noise. Several such dot pattern generator threshold matrices have been created automatically by using genetic algorithm optimization, a non-deterministic global optimization method imitating natural evolution and genetics. A hybrid of genetic algorithm with a search method based on local backtracking was developed together with several fitness functions evaluating dot patterns for rectangular grids. By modifying the fitness function, a family of dot generators results, each with its particular statistical features. Several versions of genetic algorithms, backtracking and fitness functions were tested to find a reasonable combination. The generated threshold matrices have been tested by simulating a set of test images using the Khoros image processing system. Even though the work was focused on developing low resolution marking technology, the resulting family of dot generators can be applied also in other halftoning application areas including high resolution printing technology.

Soft Computing Methods

Soft computing methods of modelling usually include fuzzy logics , neural computation , genetical algorithms and probabilistic deduction , with the addition of data mining and chaos theory in some cases. Unlike the traditional “hardcore methods” of modelling, these new methods allow for the gained results to be incomplete or inexact. Methodologically, the different approaches of these soft methods are quite heterogeneous. Still, all of them have a few things in common, namely that they have all been developed during the last 30–50 years (Bayes formula in 1763 and Lukasiewicz logic in 1920 being the exceptions), and that they would probably have not achieved their current standards without the exceptional growth in computational capacities of computers.