...482 IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART B: CYBERNETICS, VOL. 42, NO. 2, APRIL 2012 An Adaptive Differential Evolution Algorithm With Novel Mutation and Crossover Strategies for Global Numerical Optimization Sk. Minhazul Islam, Swagatam Das, Member, IEEE, Saurav Ghosh, Subhrajit Roy, and Ponnuthurai Nagaratnam Suganthan, Senior Member, IEEE Abstract—Differential evolution (DE) is one of the most powerful stochastic real parameter optimizers of current interest. In this paper, we propose a new mutation strategy, a fitnessinduced parent selection scheme for the binomial crossover of DE, and a simple but effective scheme of adapting two of its most important control parameters with an objective of achieving improved performance. The new mutation operator, which we call DE/current-to-gr_best/1, is a variant of the classical DE/current-to-best/1 scheme. It uses the best of a group (whose size is q% of the population size) of randomly selected solutions from current generation to perturb the parent (target) vector, unlike DE/current-to-best/1 that always picks the best vector of the entire population to perturb the target vector. In our modified framework of recombination, a biased parent selection scheme has been incorporated by letting each mutant undergo the usual binomial crossover with one of the p top-ranked individuals from the current population and not with the target vector with the same index as used in all variants of DE. A DE variant obtained...
Words: 11062 - Pages: 45
...144 Peter Road, Monash University south Africa 144 Peter Road, Monash University south Africa FIT 3002 Edson Zandamela and Mbuto Carlos Machili Assignment 2 Report FIT 3002 Edson Zandamela and Mbuto Carlos Machili Assignment 2 Report 08 Fall 08 Fall Table of Contents 1. Introduction 2 2. Problem Definition 2 2.1 Objective 2 2.2 Data Characteristics 2 2.3 Model Evaluation Method 3 2.4 Budgetary Constraints 3 2.5 Response rate without a model 3 3. Data Preparation and Pre-processing 4 3.1 File formatting 4 3.2 Missing Values 4 4. Experiments 4 4.1 Learning Algorithm Selection 4 4.2 Iteration Process 6 4.2.1. Attribute selection: 6 4.2.2. Changing Parameter settings 7 4.2.3. Data Normalization 7 4.2.4. Model Recommendation 8 4.2.4.1 Lift Chart 8 4.2.4.2 Gain Chart 9 5. Campaign suggestions 10 6. Conclusion 12 1. Introduction Global Paper’s prime objective is to analyze and evaluate the market response rate of a new paper product that they are currently exploring by testing the market using a mass mailing campaign. The evaluation is based on how much the product will appeal to people based on their earned salaries (<=$50k, or >$50k) per year. The company has purchased demographic data sets (Adult data set and test data) from a known source, and through market research, it has discovered that the new product is likely to appeal to persons who make over $50K a year. This report documents the data mining processes (using...
Words: 3639 - Pages: 15
...Abstract. The speech signal enhancement is needed to obtain clean speech signal from noisy signal. For multimodal optimization we better to use natural-inspired algorithms such as Firefly Algorithm (FA). We compare the firefly algorithm with particle swarm optimization technique. The proposed algorithm contains three module techniques. Those are preprocessing module, optimization module and spectral filtering module. The signals are taken from Loizou’s database and Aurora database for evaluating proposed technique. In this paper we calculate the perceptional evolution of speech quality (PESQ) and signal to noise (SNR) of the enhanced signal. The results of firefly algorithm and PSO are to be compare then we observe that the proposed technique...
Words: 887 - Pages: 4
...for Multiagent Constraint Optimization Adrian Petcu and Boi Faltings {adrian.petcu, boi.faltings}@epfl.ch http://liawww.epfl.ch/ Artificial Intelligence Laboratory Ecole Polytechnique F´ d´ rale de Lausanne (EPFL) e e IN (Ecublens), CH-1015 Lausanne, Switzerland Abstract We present in this paper a new, complete method for distributed constraint optimization, based on dynamic programming. It is a utility propagation method, inspired by the sum-product algorithm, which is correct only for tree-shaped constraint networks. In this paper, we show how to extend that algorithm to arbitrary topologies using a pseudotree arrangement of the problem graph. Our algorithm requires a linear number of messages, whose maximal size depends on the induced width along the particular pseudotree chosen. We compare our algorithm with backtracking algorithms, and present experimental results. For some problem types we report orders of magnitude fewer messages, and the ability to deal with arbitrarily large problems. Our algorithm is formulated for optimization problems, but can be easily applied to satisfaction problems as well. 1 Introduction Distributed Constraint Satisfaction (DisCSP) was first studied by Yokoo [Yokoo et al., 1992] and has recently attracted increasing interest. In distributed constraint satisfaction each variable and constraint is owned by an agent. Systematic search algorithms for solving DisCSP are generally derived from depth-first search algorithms based on some form of backtracking...
Words: 5452 - Pages: 22
... would like to tell you about heuristics and how it is used to access your database information. This report explains the implementation of an algorithm to optimize a QT with heuristic optimization rules. These rules were taken from [1] chapter 16 and [2] chapter 11. Heuristic optimization rules are based on properties of operations as mathematical operations in the relational algebra. Summaries of these properties can be found both in [1] and [2] also. These properties give the following heuristic rules for query optimization: 1. Perform SELECT operations as far down the tree as possible. This has the effect of reducing the number of tuples in later binary operations which are highly expensive. 2. Perform PROJECT operations as far down the tree as possible. This has the effect of reducing the number of attributes in each tuple and reduces the memory requirements, attempting to cut down on secondary storage usage. 3. Combine successive SELECT operations into one composite SELECT operation and successive PROJECT operations into one composite PROJECT operation. 4. Combine a PRODUCT followed by a SELECT into a JOIN with the selection condition in the SELECT. Also, combine a JOIN followed by a SELECT into a new JOIN which incorporates the selection condition. The remainder of this report describes an implementation of an algorithm which, given a suitable representation of a QT, optimizes according to the...
Words: 395 - Pages: 2
... Applied Mathematics and Computation journal homepage: www.elsevier.com/locate/amc A comparative study of Artificial Bee Colony algorithm Dervis Karaboga *, Bahriye Akay Erciyes University, The Department of Computer Engineering, Melikgazi, 38039 Kayseri, Turkey a r t i c l e i n f o a b s t r a c t Artificial Bee Colony (ABC) algorithm is one of the most recently introduced swarm-based algorithms. ABC simulates the intelligent foraging behaviour of a honeybee swarm. In this work, ABC is used for optimizing a large set of numerical test functions and the results produced by ABC algorithm are compared with the results obtained by genetic algorithm, particle swarm optimization algorithm, differential evolution algorithm and evolution strategies. Results show that the performance of the ABC is better than or similar to those of other population-based algorithms with the advantage of employing fewer control parameters. Ó 2009 Elsevier Inc. All rights reserved. Keywords: Swarm intelligence Evolution strategies Genetic algorithms Differential evolution Particle swarm optimization Artificial Bee Colony algorithm Unconstrained optimization 1. Introduction Population-based optimization algorithms find near-optimal solutions to the difficult optimization problems by motivation from nature. A common feature of all population-based algorithms is that the population consisting of possible solutions to the problem is modified by applying some operators on the solutions depending...
Words: 19739 - Pages: 79
...R Tools for Portfolio Optimization Guy Yollin Quantitative Research Analyst Rotella Capital Management Bellevue, Washington Backgrounder Rotella Capital Management Quantitative Research Analyst Systematic CTA hedge fund trading 80+ global futures and foreign exchange markets Insightful Corporation Director of Financial Engineering Developers of S-PLUS®, S+FinMetrics®, and S+NuOPT® J.E. Moody, LLC Financial Engineer Futures Trading, Risk Management, Business Development OGI School of Engineering at Oregon Health & Science University Adjunct Instructor Statistical Computing & Financial Time Series Analysis Electro Scientific Industries, Inc Director of Engineering, Vision Products Division Machine Vision and Pattern Recognition Started Using R in 1999 R Tools for Portfolio Optimization 2 Introduction DJIA: 12/02/2008 - 04/15/2009 100 GM C 80 IBM annualized return (%) JPM 60 INTC HD DD GE PG BAC R-SIG-FINANCE QUESTION: stock price 90 40 MMM 20 KFT XOM PFE CVX JNJ VZ HPQ MCD KO UTX WMT T CAT MRK AXP BA Can I do < fill in the blank > portfolio optimization in R? MSFT DIS AA 0 IBM: 12/02/2008 - 04/15/2009 0 100 5 10 conditional value-at-risk (%) 15 20 95 80 85 Maximum Drawdown Jan Mar ANSWER: drawdown (%) IBM Underwater Graph 0 -5 P/L Distribution -10 -15 Yes! (98% confidence level) 0.12 0.14 Jan Mar 0.10 VaR 0.08 Density CVaR...
Words: 663 - Pages: 3
...Scott Clark Graduate Student, DOE Computational Science Graduate Fellow 657 Rhodes Hall, Ithaca, NY, 14853 September 19, 2011 sc932@cornell.edu cam.cornell.edu/∼sc932 Education Cornell University Ph.D. Applied Math (current), M.S. Computer Science Ithaca, NY 2008 - 2012(projected) • – Department of Energy Computational Science Graduate Fellow (Full Scholarship, 4 years) – Emphasis on machine learning/data mining and algorithm design/software development related to bioinformatics and optimization • Oregon State University B.Sc. Mathematics, B.Sc. Computational Physics, B.Sc. Physics Corvallis, OR 2004 - 2008 – Graduated Magna Cum Laude with minors in Actuarial Sciences and Mathematical Sciences – Strong emphasis on scientific computing, numerical analysis and software development Skills • Development: C/C++, Python, CUDA, JavaScript, Ruby (Rails), Java, FORTRAN, MATLAB • Numerical Analysis: Optimization, Linear Algebra, ODEs, PDEs, Monte Carlo, Computational Physics, Complex Systems, Iterative Methods, Tomology • Computer Science: Machine Learning, Data Mining, Parallel Programming, Data Structures, Artificial Intelligence, Operating Systems • Discovering and implementing new ideas. Give me an API and a problem and I will figure it out. • Diverse background in Math, Computer Science, Physics and Biology allows me to communicate to a wide scientific and general audience and begin contributing to any group immediately. • I have worked in many places in a myriad of...
Words: 673 - Pages: 3
...Abbreviated version of this report is published as "Trends in Computer Science Research" Apirak Hoonlor, Boleslaw K. Szymanski and M. Zaki, Communications of the ACM, 56(10), Oct. 2013, pp.74-83 An Evolution of Computer Science Research∗ Apirak Hoonlor, Boleslaw K. Szymanski, Mohammed J. Zaki, and James Thompson Abstract Over the past two decades, Computer Science (CS) has continued to grow as a research field. There are several studies that examine trends and emerging topics in CS research or the impact of papers on the field. In contrast, in this article, we take a closer look at the entire CS research in the past two decades by analyzing the data on publications in the ACM Digital Library and IEEE Xplore, and the grants awarded by the National Science Foundation (NSF). We identify trends, bursty topics, and interesting inter-relationships between NSF awards and CS publications, finding, for example, that if an uncommonly high frequency of a specific topic is observed in publications, the funding for this topic is usually increased. We also analyze CS researchers and communities, finding that only a small fraction of authors attribute their work to the same research area for a long period of time, reflecting for instance the emphasis on novelty (use of new keywords) and typical academic research teams (with core faculty and more rapid turnover of students and postdocs). Finally, our work highlights the dynamic research landscape in CS, with its focus constantly ...
Words: 15250 - Pages: 61
...Steepest Descent Direction in Optimization - Application and Algorithm Hemanand. T Department of Chemical Engineering, St. Joseph’s College of Engineering, Chennai – 600 119 Abstract: An analytical solution to identify the minimum value of a function used for optimization based on steepest descent technique was extensively discussed with applications in a process. The properties of gradient vector, the oscillation of function values and overshoot were analyzed in a function for the search of minimum. The best step size in each iteration was found by conducting a one-D optimization in the steepest descent direction. The five steps in the algorithm for steepest descent direction were done for the effective search for the minimum included (i) estimate of a starting design and set the iteration counter, (ii) selection of a convergence parameter, calculation of the gradient of function f(x) at the point, (iii) then stop the iteration process at the minimum point, otherwise, search for minimum by next iteration, (iv) calculation of step size to minimize and (v) updation of the design with the new values which yield minimization of an optimization process. Keywords: Gradient Vector, Overshoot, One-D optimization, Convergence Parameter, Oscillation. 1. Introduction: In mathematics, optimization, or mathematical programming, refers to choosing the best element from some set of available alternatives. In the simplest case, this means solving problems in which one...
Words: 1260 - Pages: 6
...Linear Programming: Using Solver in Excel Linear Programming was conceptually developed before World War II by the outstanding Russian mathematician A.N.Kolmogorov and gained its popularity ever since the development of Simplex method by George B. Dantzig in 1947. Linear programming deals with problems of maximizing or minimizing a linear function in the presence of linear equality and/or inequality constraints. In these problems, we find the optimal, or most efficient way of using limited resources to achieve the objective of the situation. Linear Programming enables users to model large and complex problems and solve in a short amount of time by the use of effective algorithm, hence it is a powerful and widely used tool in various fields such as science, industrial engineering, financial planning and management decision making. Nowadays, with the development of technology, most of the real world Linear Programming problems are solved by computer programs. Excel Solver is a popular one. We work through different examples to demonstrate the applications of linear Programming model and the use of Excel Solver for various decision making in operation and supply chain management. Components of Linear Programming model To solve the linear programming problems, we first need to formulate the mathematical description called a mathematical model to represent the situation. Linear programming model usually consists of the following components * Decision variables: These represent...
Words: 2395 - Pages: 10
...problem in large scale RFID systems : Algorithms, performance evaluation and discussions John Sum, Kevin Ho, Siu-chung Lau Abstract—Assigning neighboring RFID readers with nonoverlapping interrogation time slots is one approach to solve the reader collision problem. In which, Distributed Color Selection (DCS) and Colorwave algorithm have been developed, and simulated annealing (SA) technique have been applied. Some of them (we call them non-progresive algorithms), like DCS, require the user to pre-defined the number of time slots. While some of them (we call them progressive), like Colorwave, determine the number automatically. In this paper, a comparative analysis on both non-progressive and progressive algorithms to solve such a problem in a random RFID reader network is presented. By extensive simulations on a dense network consisting of 250 readers whose transmission rates are 100%, a number of useful results have been found. For those non-progressive type algorithms, it is found that DCS is unlikely to generate a collision-free solution, even the number of time slots is set to 20. On the other hand, heuristic and SAbased algorithms can produce collision-free solutions whenever the number of time slots is set to 16. For the cases when the number of time slots is not specified, heuristic-based, SAbased and Colorwave algorithms are all able to determine the number automatically and thus generate collision-free solution. However, SA-based algorithms require much longer time than the...
Words: 6608 - Pages: 27
...Software and deployment optimization Algorithmic efficiency Main article: Algorithmic efficiency Further information: Analysis of algorithms The efficiency of algorithms has an impact on the amount of computer resources required for any given computing function and there are many efficiency trade-offs in writing programs. Algorithm changes, such as switching from a slow (e.g. linear) search algorithm to a fast (e.g. hashed or indexed) search algorithm can reduce resource usage for a given task from substantial to close to zero. A study by a physicist at Harvard, estimated that the average Google search released 7 grams of carbon dioxide (CO₂).[25] However, Google disputes this figure, arguing instead that a typical search produces only 0.2 grams of CO₂.[26] Resource allocation Main article: Resource allocation Algorithms can also be used to route data to data centers where electricity is less expensive. Researchers from MIT, Carnegie Mellon University, and Akamai have tested an energy allocation algorithm that successfully routes traffic to the location with the cheapest energy costs. The researchers project up to a 40 percent savings on energy costs if their proposed algorithm were to be deployed. However, this approach does not actually reduce the amount of energy being used; it reduces only the cost to the company using it. Nonetheless, a similar strategy could be used to direct traffic to rely on energy that is produced in a more environmentally friendly or efficient...
Words: 370 - Pages: 2
...cameras capture, what the people it’s observing are doing,” explains Dr. Raymond Perrault, director, Artificial Intelligence Center, SRI. “It can tell you that someone is walking through the door, or that two people have met and exchanged a package or that a person is digging a hole by the side of the road.” A robot like this with sensors could decide what information is pertinent and report the data to a group of warfighters. These intelligence systems can perceive their environment and adjust. “They manage to do [their mission] while the world changes around them,” Perrault says. To accomplish this task, they organize ideas utilizing mathematical logic. Using sensory data, the programs prove simple theorems by plugging the data into the algorithms, which results in a solution and consequent action.” Garegnani, J. (2010, December 15) Artificial Software is good for Aerospace because it gives us the capabilities to do more traveling into space than ever before. Intelligent System (IS) applications have gained popularity among aerospace professionals in the last decade due to the ease with which several of the IS tools can be...
Words: 564 - Pages: 3
...Credit Prerequisites Language ONUR KAYA W 14:00-16:00 ENG 206 1583 okaya@ku.edu.tr 3 6 INDR. 262 English Assistant TA/RA/Lab Assistant Name AYLİN LELİZAR POLAT GÜLÇİN ERMİŞ Email aypolat@ku.edu.tr gulermis@ku.edu.tr Office Hours Office Location Course Description Introduction to modeling with integer variables and integer programming; network models, dynamic programming; convexity and nonlinear optimization; applications of various optimization methods in manufacturing, product design, communications networks, transportation, supply chain, and financial systems. Course Objectives The course is designed to teach the concepts of optimization models and solution methods that include integer variables and nonlinear constraints. Network models, integer, dynamic and nonlinear programming will be introduced to the students. Students will be exposed to applications of various optimization methods in manufacturing, product design, communications networks, transportation, supply chain, and financial systems. Several different types of algorithms will also be presented to solve these problems. The course also aims to teach how to use computer programs such as Matlab and GAMS to solve mathematical models. Learning Outcomes Students are expected to model real life problems using mathematical models including integer variables and nonlinear equations. Students will be able to apply mathematical modeling techniques such as dynamic, integer and nonlinear programming...
Words: 768 - Pages: 4