DIFFERENTIAL evolution (DE) [3] has become a popular

Size: px
Start display at page:

Download "DIFFERENTIAL evolution (DE) [3] has become a popular"

Transcription

1 Self-adative Differential Evolution with Neighborhood Search Zhenyu Yang, Ke Tang and Xin Yao Abstract In this aer we investigate several self-adative mechanisms to imrove our revious work on [], which is a recent DE variant for numerical otimization. The selfadative methods originate from another DE variant, [2], but are remarkably modified and extended to fit our. And thus a Self-adative (Sa) is roosed to imrove s erformance. Three self-adative mechanisms are utilized in Sa: self-adatation for two candidate mutation strategies, self-adatations for controlling scale factor F and crossover rate CR, resectively. Exerimental studies are carried out on a broad range of different benchmark functions, and the roosed Sa has shown significant sueriority over. I. INTRODUCTION DIFFERENTIAL evolution (DE) [3] has become a oular algorithm in global otimization. It has shown suerior erformance in both widely used benchmark functions [4], [5] and real-world alications [6]. DE conventionally has several candidate mutation schemes, and three control arameters, i.e., oulation size NP, scale factor F and crossover rate CR. Aart from the arameter NP (which is common for all oulation-based algorithms), mutation strategy selection, arameters F and CR adatations are the three most imortant issues of DE research. Many work has been done along these lines. The relationshi between the control arameters and oulation diversity has been analyzed in [7]. Exerimental arameter studies and emirical arameter settings of DE have been carried out in [8]. Selfadative strategy has also been investigated to adat these control arameters [9], as well as different mutation strategies [2]. In [], we roosed a DE variant, namely Differential Evolution with Neighborhood Search (), to adat the scale factor F. Insired by the neighborhood search (NS) strategy in evolutionary rogramming (EP) [], intends to mix search biases of different NS oerators through the factor F. It is well-known that NS is a main strategy underinning EP []. Although DE might be similar to the evolutionary rocess of EP, it lacks relevant concet of neighborhood search. Instead of redefining the factor F as a constant, generates F from Gaussian and Cauchy distributed random numbers, which are beneficial to roducing small and large search ste sizes, resectively []. A robability is introduced to control when to use Gaussian or Cauchy oerator. In the revious work was simly set to a The authors are with the Nature Insired Comutation and Alications Laboratory, the Deartment of Comuter Science and Technology, University of Science and Technology of China, Hefei, Anhui 2327, China. Xin Yao is also with CERCIA, the School of Comuter Science, University of Birmingham, Edgbaston, Birmingham B5 2TT, U.K. ( s: zhyuyang@mail.ustc.edu.cn, ketang@ustc.edu.cn, x.yao@cs.bham.ac.uk). Corresonding author: Ke Tang (Phone: ). constant number. Obviously, it would be more desirable if could be self-adated during the evolution rocess. By these means, the algorithm can automatically adjust between Gaussian and Cauchy oerators, and thereby the erformance can be imroved. Self-adative Differential Evolution () by Qin et at. [2], is a different DE variant that mainly focuses on adatation for arameter CR and mutation strategies of DE. The motivation is to solve the dilemma that CR values and mutation strategies involved in DE are often highly roblem deendent. adots two DE mutation strategies and introduces a robability to control which one to use. The robability is gradually self-adated according to learning exerience. Additionally, crossover rate CR is self-adated by recording CR values that make offsring successfully enter the next generation. Both of the two self-adative mechanisms have achieved significant imrovement over the classical DE with emirical arameter configuration. It can be concluded that and have quite different emhases on imroving DE s erformance. ays secial attention to the crossover rate CR s adatation and the self-adatation between different DE mutation strategies, while intends to mix search biases of different NS oerators through the arameter F, and no self-adatation is adoted. The difference motivates us to introduce s self-adative mechanisms into, study their behaviors, and then roose a self-adative (Sa). The outline and features of the roosed Sa are summarized as follows: ) It inherits the self-adated mutation schemes selection mechanism of ; 2) It adots a selfadative strategy to adjust the arameter of ; 3) It enhances the original CR self-adatation of by adding a weighting strategy. The efficacy of Sa is evaluated on two sets of widely used benchmark functions. The rest of this aer is organized as follows: Section II gives the reliminaries; Section III describes the roosed Sa algorithm; Section IV resents the exerimental studies; Finally, Section V concludes this aer and briefly discusses several other self-adative DE schemes. II. PRELIMINARIES A. Differential Evolution (DE) Individuals in DE are reresented by D-dimensional vectors x i, i,, NP}, whered is the number of objective arameters and NP is the oulation size. According to [3], the classical DE can be summarized as follows: ) Mutation: v i = x i + F (x i2 x i3 ) /8/$25. c 28 IEEE

2 where i,i 2,i 3 [, NP] are random and mutually different integers, and they are also different with the vector index i. Scale factor F > is a real constant factor and is often set to. 2) Crossover: vi (j), if U u i (j) = j (, ) CR or j = j rand x i (j), otherwise. with U j (, ) stands for the uniform random number between and, and j rand is a randomly chosen index to ensure that the trial vector u i does not dulicate x i. CR (, ) is the crossover rate, which is often set to.9. 3) Selection: x i = ui, if f(u i ) f(x i ) x i, otherwise. where x i is the offsring of x i for the next generation (Without loss of generality, we consider only minimization roblem in this aer). There are several schemes of DE based on different mutation strategies [3]: v i = x i + F (x i2 x i3 ) () v i = x best + F (x i x i2 ) (2) v i = x i + F (x best x i )+F (x i x i2 ) (3) v i = x best + F (x i x i2 )+F (x i3 x i4 ) (4) v i = x i + F (x i2 x i3 )+F (x i4 x i5 ) (5) Schemes () and (3), with notations as DE/rand/ and DE/current to best/2, are the most often used in ractice due to their good erformance [2], [3]. B. Differential Evolution with Neighborhood Search () [] is a recent DE variant that utilizes the neighborhood search (NS) strategy in evolutionary rogramming (EP). NS is a main strategy underinning EP, and the characteristics of several NS oerators have been investigated in EP literature []. Although DE might be similar to the evolutionary rocess in EP, it lacks relevant concet of neighborhood search. is the same with the classical DE described in Section II.A excet the scale factor F is relaced by the following equation: Ni (, ), if U F i = i (, ) < (6) δ i, otherwise. where i is the index of current trial vector, U i (, ) stands for the uniform random number between and, N i (, ) denotes a Gaussian random number with mean and standard deviation, and δ i denotes a Cauchy random variable with scale arameter t =. The arameter was set to a constant number in. The advantages of NS strategy in DE have been studied in []. Exerimental results have shown that has significant advantages over classical DE on a broad range of different benchmark functions. It has been found that is effective in escaing from local otima when searching in environments without rior knowledge about what kind of search ste size will be referred. C. Self-adative Differential Evolution () by Qin et al. [2], gives the first attemt to adot two different mutation strategies in single DE variant. The motivation of is to solve the dilemma that mutation strategies involved in DE are often highly deendent on the roblems under consideration. It introduces a robability to control which mutation strategy to use, and is gradually self-adated according to the learning exerience. Additionally, utilizes two methods to adat and self-adat DE s arameters F and CR. Detailed contributions of are summarized as follows: ) Mutation strategies self-adatation: selects mutation strategies Eq. () and Eq. (3) as candidates, and roduces the trial vector based on: Eq. (), if Ui (, ) < v i = (7) Eq. (3), otherwise. Here is set to initially. After evaluation of all offsring, the number of offsring successfully entering the next generation while generated by Eq. () and Eq. (3) are recorded as ns and ns 2, resectively, and the numbers of offsring discarded while generated by Eq. () and Eq. (3) are recorded as nf and nf 2. Those two airs of numbers are accumulated within a secified number of generations (5 in ), called the learning eriod. Then, the robability is udated as: ns (ns 2 + nf 2 ) = (8) ns 2 (ns + nf )+ns (ns 2 + nf 2 ) Here ns, ns 2, nf and nf 2 will be reset once is udated after each learning eriod. 2) Scale factor F setting: In, F is set to F i = N i (,.3) where N i (,.3) denotes a Gaussian random number with mean and standard deviation.3. 3) Crossover rate CR self-adatation: allocates a CR i for each individuals according to: CR i = N i (,.) (9) is set to initially. These CR values for all individuals remain the same for several generations (5 in ) and then a new set of CR values is generated using the same equation. During every generation, the CR values associated with offsring successfully entering the next generation are recorded in an array CR rec. After a secified number of generations (25 in ), will be udated: = CR rec CR rec (k) () CR rec k= 28 IEEE Congress onevolutionarycomutation(cec 28)

3 CR rec will be reset once is udated. This selfadatation scheme for CR is denoted as SaCR. For detailed rinciles and exlanations behind s selfadatation strategies, arameter settings, or even simulated results, lease refer to [2]. III. SELF-ADAPTIVE DIFFERENTIAL EVOLUTION WITH NEIGHBORHOOD SEARCH A. Sa: The Incororated Algorithms It can be concluded that and have quite different emhases: The former ays secial attention to selfadatation between different mutation strategies, as well as the self-adatation on crossover rate CR, while the latter intends to mix search biases of different NS oerators through the arameter F, and no self-adatation is adoted. The difference motivates us to introduce s self-adative mechanisms into, study their behaviors, and then roose a self-adative (Sa). Based on the motivations above, we address crucial issues of the roosed Sa as follows: ) Mutation strategies self-adatation: Sa utilizes the same method as in this art. For details, lease refer to Eq. (7) and Eq. (8). 2) Scale factor F self-adatation: Sa inherits the method of controlling the arameter F from, but extending it to: Ni (,.3), if U F i = i (, ) < δ i, otherwise. where will be self-adated as is done in according to Eq. (8), excet here we have to record corresonding F values that make offsring enter the next generation successfully. 3) Weighted crossover rate CR self-adatation: We use a similar strategy to what does with SaCR strategy. But whenever we record a successful CR value in array CR rec, we will also record the corresonding imrovement on fitness value in array Δf rec, with Δf rec (k) =f(k) f new (k). And then, Eq. () is changed to: = CR rec k= w k =Δf rec (k)/ w k CR rec (k) () Δf rec k= Δf rec (k) (2) Note: here CR rec Δf rec. The weighted selfadatation scheme for CR is denoted as SaCRW, and we will exlain why we add the weight mechanism to the original SaCR in Section III.B with details. Due to the significant successes of and, Sa, which incororates enhancements ), 2) and 3), is romising. B. Weighted CR Self-adatation The arameter CR of DE determines how many comonents of mutated vector will be introduced into current candidate for the next generation, so the robability of generating imroved offsring from the same arent with a small CR is higher than that with a large CR. It can referred that the of SaCR has an imlicit bias towards small values during self-adatation rocess. The bias might become harmful when otimizing nonsearable functions, in which interactions exist between variables. Because large CR value is required to change the nonsearable variables together. To illustrate this roblem, we conducted an exeriment with Sa+SaCR on the well-known Generalized Rosenbrock s function []. The evolution curves for S runs and F runs of 25 indeendent runs are given in Fig.. Here S runs means Sa has found the region of otimum, while F runs means Sa failed to do that. For S runs, it can be found that was successfully adated to a large value, and after that the fitness values are imroved quickly. For F runs, was adated to a small value, and traed there from that time on. The algorithm failed to make significant imrovement on fitness values thereafter..2 2 S runs F runs 2 S runs F runs 4 fitness value Fig.. Evolution curves of and fitness value on the Generalized Rosenbrock function. S runs denotes results of successful runs, while F runs denotes results of failed runs of 25 indeendent runs. The vertical axes show the value (u figure) and fitness value (down figure), while the horizontal axes show the number of generations. On the other hand, it is assumed that large successful CR values will achieve larger imrovement on fitness values than small successful CR values for nonsearable functions, because it will be good to change nonsearable variables together []. So we can balance the bias of SaCR with a weight based on the size of fitness value imrovement. This is the basic motivation of SaCRW in Section III.A. To validate the effectiveness, another exeriment with Sa+SaCRW was conducted on the same function. The results are summarized in Table I. It can be found that SaCRW was successfully adats to required large values in all runs. The advantage is also shown by differences of fitness values IEEE Congress onevolutionarycomutation(cec 28)

4 TABLE I SIMULATED RESULTS OF SACR AND SACRW ON THE GENERALIZED ROSENBROCK FUNCTION.THE RESULTS OF 25 INDEPENDENT RUNS ARE SORTED FROM ST TO 25TH BASED ON FITNESS VALUES # of Sa+SaCR Sa+SaCRW runs Fitness Final Fitness Final st 7.42e-3 4.e+ 88 5th 3.99e+.5.e+ 34 9th 2.33e+.64.e+ 53 3th 2.34e+.57.e+ 73 7th 2.37e e th 2.39e e th 2.43e+.55.9e Mean.82e+ 4.3e-3 Std 9.76e+ 6.28e-3 SaCRW made the algorithms success in all 25 runs, while SaCR made it achieve only 4 successful runs (st 4th). IV. EXPERIMENTAL STUDIES A. Exerimental Setu Exerimental validations for the roosed Sa are conducted on both a set of classical test functions [], and a new set of benchmark functions rovided by CEC 25 secial session [2]. The algorithms used for comarison are, and Sa (with SaCRW). The oulation size NP is set to for all algorithms, and no other arameters is adjusted during evolution. B. Results on Classical Benchmark Functions The classical test set includes 23 functions, in which f f 3 are high-dimensional (3-D) and f 4 f 23 are low-dimensional functions. Functions f f 5 are unimodal, functions f 8 f 3 are multimodal functions with many local otima, and functions f 4 f 23 are multimodal functions with only a few local otima. Details of these functions can be found in the aendix of []. The number of evolution generations of all algorithms is set to 5 for f f 4, 5 for f 5, 5 for f 6 f 3, 2 for f 4, 5 for f 5 and 2 for f 6 f 23. The average results of 25 indeendent runs are summarized in Tables II IV. TABLE II EXPERIMENTAL COMPARISON ON f f 7 (OVER 25 RUNS). Func Mean Mean Mean t-test t-test f 3.2e e e f e- 6.22e- 4.5e f e-22.2e-8.6e f 4.59e e e f 5 4.3e-3 2.e+.24e f 6.e+.e+.e+.. f 7 7.2e e-3.2e For unimodal functions f f 7, Sa achieved much better results than and, excet on the simle ste function f 6, where all three algorithms erformed exactly the same. The great difference can be seen from results on the Generalized Rosenbrock s Function, f 5. The evolution curves of arameter adatation and fitness value for this function are given in Fig. 2 and 3. As we mentioned before, in Sa has been able to self-adat to roer values. TABLE III EXPERIMENTAL COMPARISON ON f 8 f 3 (OVER 25 RUNS). Func Mean Mean Mean t-test t-test f f 9.84e-5 4.e e f 2.36e-2 9.6e- 6.72e f.e+ 8.88e e f e-23.2e e f 3 3.2e-22.75e e For multimodal functions f 8 f 3, Sa is the clear winner again, excet that it was outerformed by on function f 9. With further observation of curves on Figs. 2 and 3, Sa converged slower than, but still made good imrovement all the way. This might have haened because the self-adated arameters in Sa need more time to find the roer values on this function. TABLE IV EXPERIMENTAL COMPARISON ON f 4 f 23 (OVER 25 RUNS). Func Mean Mean Mean t-test t-test f f 5 3.7e-4 3.7e-4 3.7e-4.. f f f 8 3.e+ 3.e+ 3.e+.. f f f f f Table IV shows the results for low-dimensional functions f 4 f 23. The three comared algorithms showed only very minor differences on f 2 and f 2 f 23 (which cannot be seen from the mean values). That is because all of the algorithms have suerior erformance on this low-dimensional functions. C. Results on CEC 25 Benchmark Functions To evaluate Sa further, a new set of benchmark functions rovided by CEC 25 secial session was used. It includes 25 functions with different comlexity [2]. Functions f cec f cec5 are unimodal while the remaining 2 functions are multimodal. Since functions f cec5 f cec25 are hybrid comosition functions, which are very time consuming for fitness evaluation, we only used the first 4 functions of the set in our exeriments. All of these functions are scalable, and we set their dimensions to 3 in our exeriments. Detailed descrition of these functions can be found in [2]. The number of evolution generations is set to 3 for all 28 IEEE Congress onevolutionarycomutation(cec 28) 3

5 5 Sa.2.7 f f Sa f f Sa f cec5 f cec Sa f cec9 2 f cec9 Fig. 2. The self-adatation curves of, and for f 5, f 9, f cec5, and f cec9. On the vertical axes are shown their values (between and ), while on the horizontal axes are shown the number of generations. Fig. 3. The evolution curves for f 5, f 9, f cec5 and f cec9. The vertical axes show the distance to the otimum and the horizontal axes show the number of generations IEEE Congress onevolutionarycomutation(cec 28)

6 functions. Error value, i.e. the difference between current fitness value and otimum, is used to comare algorithm s erformance. The average error values of 25 indeendent runs are summarized in Tables V and VI. TABLE V EXPERIMENTAL COMPARISON ON f cec f cec5 (OVER 25 RUNS). Func Error Error Error t-test t-test f cec.e+.e+.e+.. f cec2 5.68e-4.25e-3 4.e f cec3 5.43e+4.77e+5.67e f cec4.22e-4.89e e f cec5 2.45e-.e+3.5e For unimodal functions, Sa erformed better than the other two algorithms on all 5 functions, and significantly better on functions f cec2, f cec4 and f cec5. This is consistent with conclusions drawn on classical unimodal functions. The effectiveness and efficiency on f cec5 can be seen from evolution curves in Fig. 2 and 3. TABLE VI EXPERIMENTAL COMPARISON ON f cec6 f cec4 (OVER 25 RUNS). Func Error Error Error t-test t-test f cec6.59e- 2.99e+ 2.89e f cec7 8.57e-3.65e-2.2e f cec8 2.9e+ 2.9e+ 2.9e+.9.8 f cec9.e+ 2.27e-5.99e f cec 4.2e+ 5.5e+ 4.24e f cec.2e+ 2.7e+.48e f cec2 4.6e e+4.74e f cec3 2.2e+ 2.e+ 5.e f cec4.27e+.26e+.32e For multimodal functions, Sa obtained better results on almost all functions, excet on f cec3 and f cec4,whereit was outerformed by. All algorithms erformed badly on the two functions, and haened to show a minor sueriority. f cec3 and f cec4 are exanded functions, which are comosed of other different functions. This makes their characteristics unclear for further case study. The analysis of algorithms evolutionary behaviors on functions like them is one of the focuses of our future work. Fig. 3 shows the evolution curves of f cec9. It can be seen that Sa found the otimum in less than the maximum number of available generations. The curves in Fig. 2 of this functions showed Sa required different values for, and during different stages, and the self-adatation strategies were able to adjust these arameters as needed (from large to small, then to large again). V. CONCLUSIONS AND DISCUSSIONS In this aer, we roosed a new self-adative DE variant, Sa, which is an imroved version of our revious algorithm. The Sa can be viewed as a hybridization of [2] and []. In Sa: ) We utilized the self-adatation strategy of to adat between candidate mutations; 2) We alied a self-adatation to adjust arameter F ; 3) We illustrated the ill-condition of original CR self-adatation in, and roosed an enhanced version with weighting. The erformance of the roosed Sa algorithm is evaluated and discussed on both a set of 23 classical test functions[], and a new set of 4 benchmark functions rovided by CEC 25 secial session [2]. Sa has shown significant sueriority over both and. Besides the mentioned in this aer, several other self-adative DE variants (s) have also been roosed. Omran et al. roosed a SDE [3], [4] by adating arameters F and CR based on normal distribution. Brest et al. resented the jde, which attaches F and CR values to all individuals of oulation, and evolves these control arameters at individual level [9]. F and CR are udated in each generation according to some heuristic rules. In their later work [5], an imroved version of jde, namely jde-2, has also been roosed by imorting the mutation strategies self-adatation from (Qin et al.). In some resects, Sa is the inheritor of and, while it is different from other self-adative DE algorithms in two major asects: ) By mixing the search biases of both Gaussian and Cauchy oerators, Sa considers a trade-off between small and large search ste sizes; 2) Sa self-adats all its control arameters according to statistical learning exerience during evolution, rather than other heuristic udating rules. We have comared the erformance of Sa with these latest s, but the lack of sace revents showing the results of those exeriments. In general, Sa achieved comarable results to that of the other methods. ACKNOWLEDGMENT This work is artially suorted by the National Natural Science Foundation of China (Grant No ), the Fund for Foreign Scholars in University Research and Teaching Programs (Grant No. B733), and the Graduate Innovation Fund of University of Science and Technology of China (Grant No. KD2744). REFERENCES [] Z. Yang, J. He, and X. Yao, Making a Difference to Differential Evolution, in Advances in Metaheuristics for Hard Otimization, Z. Michalewicz and P. Siarry, Eds. Sringer, 28, [2] A. K. Qin and P. N. Suganthan, Self-adative differential evolution algorithm for numerical otimization, Proceedings of the 25 IEEE Congress on Evolutionary Comutation, vol. 2, , 25. [3] K. Price, R. Storn, and J. Laminen, Differential Evolution: A Practical Aroach to Global Otimization. Sringer-Verlag, ISBN: , 25. [4] J. Vesterstrom and R. Thomsen, A comarative study of differential evolution, article swarm otimization, and evolutionary algorithms on numerical benchmark roblems, Proceedings of the 24 Congress on Evolutionary Comutation, vol. 2, , 24. [5] N. Hansen, Comilation of results on the CEC benchmark function set, Institute of Comutational Science, ETH Zurich, Switerland, Tech. Re, vol. 3, 25. [6] R. Storn, System design by constraint adatation and differential evolution, IEEE Transactions on Evolutionary Comutation, vol. 3, no., , IEEE Congress onevolutionarycomutation(cec 28) 5

7 [7] D. Zaharie, Critical Values for the Control Parameters of Differential Evolution Algorithms, Proceedings of the 8th International Conference on Soft Comuting, , 22. [8] R. Gamerle, S. Muller, and P. Koumoutsakos, A Parameter Study for Differential Evolution, Proceedings WSEAS international conference on advances in intelligent systems, fuzzy systems, evolutionary comutation, , 22. [9] J. Brest, S. Greiner, B. Boskovic, M. Mernik, and V. Žumer, Self- Adating Control Parameters in Differential Evolution: A Comarative Study on Numerical Benchmark Problems, IEEE Transactions on Evolutionary Comutation, vol. 2,. 82 2, 26. [] T. Bäck and H. P. Schwefel, An overview of evolutionary algorithms for arameter otimization, Evolutionary Comutation, vol., no.,. 23, 993. [] X. Yao, Y. Liu, and G. Lin, Evolutionary Programming Made Faster, IEEE Transactions on Evolutionary Comutation, vol. 3, no. 2,. 82 2, 999. [2] P. N. Suganthan, N. Hansen, J. J. Liang, K. Deb, Y. P. Chen, A. Auger, and S. Tiwari, Problem Definitions and Evaluation Criteria for the CEC 25 Secial Session on Real-Parameter Otimization, Technical Reort, Nanyang Technological University, Singaore, htt:// 25. [3] M. Omran, A. Salman, and A. Engelbrecht, Self-adative Differential Evolution, Proceedings of the 25 International Conference on Comutational Intelligence and Security, , 25. [4] A. Salman, A. Engelbrecht, and M. Omran, Emirical analysis of self-adative differential evolution, Euroean Journal of Oerational Research, vol. 83, no. 2, , 27. [5] J. Brest, B. Bošković, S. Greiner, V. Žumer, and M. Maučec, Performance comarison of self-adative and adative differential evolution algorithms, Soft Comuting-A Fusion of Foundations, Methodologies and Alications, vol., no. 7, , IEEE Congress onevolutionarycomutation(cec 28)

Multi-start JADE with knowledge transfer for numerical optimization

Multi-start JADE with knowledge transfer for numerical optimization Multi-start JADE with knowledge transfer for numerical optimization Fei Peng, Ke Tang,Guoliang Chen and Xin Yao Abstract JADE is a recent variant of Differential Evolution (DE) for numerical optimization,

More information

A Scalability Test for Accelerated DE Using Generalized Opposition-Based Learning

A Scalability Test for Accelerated DE Using Generalized Opposition-Based Learning 009 Ninth International Conference on Intelligent Systems Design and Applications A Scalability Test for Accelerated DE Using Generalized Opposition-Based Learning Hui Wang, Zhijian Wu, Shahryar Rahnamayan,

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

Parallel Quantum-inspired Genetic Algorithm for Combinatorial Optimization Problem

Parallel Quantum-inspired Genetic Algorithm for Combinatorial Optimization Problem Parallel Quantum-insired Genetic Algorithm for Combinatorial Otimization Problem Kuk-Hyun Han Kui-Hong Park Chi-Ho Lee Jong-Hwan Kim Det. of Electrical Engineering and Comuter Science, Korea Advanced Institute

More information

4. Score normalization technical details We now discuss the technical details of the score normalization method.

4. Score normalization technical details We now discuss the technical details of the score normalization method. SMT SCORING SYSTEM This document describes the scoring system for the Stanford Math Tournament We begin by giving an overview of the changes to scoring and a non-technical descrition of the scoring rules

More information

SIMULATED ANNEALING AND JOINT MANUFACTURING BATCH-SIZING. Ruhul SARKER. Xin YAO

SIMULATED ANNEALING AND JOINT MANUFACTURING BATCH-SIZING. Ruhul SARKER. Xin YAO Yugoslav Journal of Oerations Research 13 (003), Number, 45-59 SIMULATED ANNEALING AND JOINT MANUFACTURING BATCH-SIZING Ruhul SARKER School of Comuter Science, The University of New South Wales, ADFA,

More information

An Ant Colony Optimization Approach to the Probabilistic Traveling Salesman Problem

An Ant Colony Optimization Approach to the Probabilistic Traveling Salesman Problem An Ant Colony Otimization Aroach to the Probabilistic Traveling Salesman Problem Leonora Bianchi 1, Luca Maria Gambardella 1, and Marco Dorigo 2 1 IDSIA, Strada Cantonale Galleria 2, CH-6928 Manno, Switzerland

More information

Genetic Algorithm Based PID Optimization in Batch Process Control

Genetic Algorithm Based PID Optimization in Batch Process Control International Conference on Comuter Alications and Industrial Electronics (ICCAIE ) Genetic Algorithm Based PID Otimization in Batch Process Control.K. Tan Y.K. Chin H.J. Tham K.T.K. Teo odelling, Simulation

More information

Genetic Algorithms, Selection Schemes, and the Varying Eects of Noise. IlliGAL Report No November Department of General Engineering

Genetic Algorithms, Selection Schemes, and the Varying Eects of Noise. IlliGAL Report No November Department of General Engineering Genetic Algorithms, Selection Schemes, and the Varying Eects of Noise Brad L. Miller Det. of Comuter Science University of Illinois at Urbana-Chamaign David E. Goldberg Det. of General Engineering University

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

Adaptive Differential Evolution and Exponential Crossover

Adaptive Differential Evolution and Exponential Crossover Proceedings of the International Multiconference on Computer Science and Information Technology pp. 927 931 ISBN 978-83-60810-14-9 ISSN 1896-7094 Adaptive Differential Evolution and Exponential Crossover

More information

THE objective of global optimization is to find the

THE objective of global optimization is to find the Large Scale Global Optimization Using Differential Evolution With Self-adaptation and Cooperative Co-evolution Aleš Zamuda, Student Member, IEEE, Janez Brest, Member, IEEE, Borko Bošković, Student Member,

More information

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning

Uncorrelated Multilinear Principal Component Analysis for Unsupervised Multilinear Subspace Learning TNN-2009-P-1186.R2 1 Uncorrelated Multilinear Princial Comonent Analysis for Unsuervised Multilinear Subsace Learning Haiing Lu, K. N. Plataniotis and A. N. Venetsanooulos The Edward S. Rogers Sr. Deartment

More information

Research Article Research on Evaluation Indicator System and Methods of Food Network Marketing Performance

Research Article Research on Evaluation Indicator System and Methods of Food Network Marketing Performance Advance Journal of Food Science and Technology 7(0: 80-84 205 DOI:0.9026/afst.7.988 ISSN: 2042-4868; e-issn: 2042-4876 205 Maxwell Scientific Publication Cor. Submitted: October 7 204 Acceted: December

More information

An Analysis of Reliable Classifiers through ROC Isometrics

An Analysis of Reliable Classifiers through ROC Isometrics An Analysis of Reliable Classifiers through ROC Isometrics Stijn Vanderlooy s.vanderlooy@cs.unimaas.nl Ida G. Srinkhuizen-Kuyer kuyer@cs.unimaas.nl Evgueni N. Smirnov smirnov@cs.unimaas.nl MICC-IKAT, Universiteit

More information

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO)

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO) Combining Logistic Regression with Kriging for Maing the Risk of Occurrence of Unexloded Ordnance (UXO) H. Saito (), P. Goovaerts (), S. A. McKenna (2) Environmental and Water Resources Engineering, Deartment

More information

Estimation of component redundancy in optimal age maintenance

Estimation of component redundancy in optimal age maintenance EURO MAINTENANCE 2012, Belgrade 14-16 May 2012 Proceedings of the 21 st International Congress on Maintenance and Asset Management Estimation of comonent redundancy in otimal age maintenance Jorge ioa

More information

Discrete Particle Swarm Optimization for Optimal DG Placement in Distribution Networks

Discrete Particle Swarm Optimization for Optimal DG Placement in Distribution Networks Discrete Particle Swarm Otimization for Otimal DG Placement in Distribution Networs Panaj Kumar, Student Member, IEEE, Nihil Guta, Member, IEEE, Anil Swarnar, Member, IEEE, K. R. Niazi, Senior Member,

More information

Dynamic Optimization using Self-Adaptive Differential Evolution

Dynamic Optimization using Self-Adaptive Differential Evolution Dynamic Optimization using Self-Adaptive Differential Evolution IEEE Congress on Evolutionary Computation (IEEE CEC 2009), Trondheim, Norway, May 18-21, 2009 J. Brest, A. Zamuda, B. Bošković, M. S. Maučec,

More information

Linear diophantine equations for discrete tomography

Linear diophantine equations for discrete tomography Journal of X-Ray Science and Technology 10 001 59 66 59 IOS Press Linear diohantine euations for discrete tomograhy Yangbo Ye a,gewang b and Jiehua Zhu a a Deartment of Mathematics, The University of Iowa,

More information

Improved Identification of Nonlinear Dynamic Systems using Artificial Immune System

Improved Identification of Nonlinear Dynamic Systems using Artificial Immune System Imroved Identification of Nonlinear Dnamic Sstems using Artificial Immune Sstem Satasai Jagannath Nanda, Ganaati Panda, Senior Member IEEE and Babita Majhi Deartment of Electronics and Communication Engineering,

More information

A New GP-evolved Formulation for the Relative Permittivity of Water and Steam

A New GP-evolved Formulation for the Relative Permittivity of Water and Steam ew GP-evolved Formulation for the Relative Permittivity of Water and Steam S. V. Fogelson and W. D. Potter rtificial Intelligence Center he University of Georgia, US Contact Email ddress: sergeyf1@uga.edu

More information

DETC2003/DAC AN EFFICIENT ALGORITHM FOR CONSTRUCTING OPTIMAL DESIGN OF COMPUTER EXPERIMENTS

DETC2003/DAC AN EFFICIENT ALGORITHM FOR CONSTRUCTING OPTIMAL DESIGN OF COMPUTER EXPERIMENTS Proceedings of DETC 03 ASME 003 Design Engineering Technical Conferences and Comuters and Information in Engineering Conference Chicago, Illinois USA, Setember -6, 003 DETC003/DAC-48760 AN EFFICIENT ALGORITHM

More information

PERFORMANCE BASED DESIGN SYSTEM FOR CONCRETE MIXTURE WITH MULTI-OPTIMIZING GENETIC ALGORITHM

PERFORMANCE BASED DESIGN SYSTEM FOR CONCRETE MIXTURE WITH MULTI-OPTIMIZING GENETIC ALGORITHM PERFORMANCE BASED DESIGN SYSTEM FOR CONCRETE MIXTURE WITH MULTI-OPTIMIZING GENETIC ALGORITHM Takafumi Noguchi 1, Iei Maruyama 1 and Manabu Kanematsu 1 1 Deartment of Architecture, University of Tokyo,

More information

Feedback-error control

Feedback-error control Chater 4 Feedback-error control 4.1 Introduction This chater exlains the feedback-error (FBE) control scheme originally described by Kawato [, 87, 8]. FBE is a widely used neural network based controller

More information

Investigation of Mutation Strategies in Differential Evolution for Solving Global Optimization Problems

Investigation of Mutation Strategies in Differential Evolution for Solving Global Optimization Problems Investigation of Mutation Strategies in Differential Evolution for Solving Global Optimization Problems Miguel Leon Ortiz and Ning Xiong Mälardalen University, Västerås, SWEDEN Abstract. Differential evolution

More information

AI*IA 2003 Fusion of Multiple Pattern Classifiers PART III

AI*IA 2003 Fusion of Multiple Pattern Classifiers PART III AI*IA 23 Fusion of Multile Pattern Classifiers PART III AI*IA 23 Tutorial on Fusion of Multile Pattern Classifiers by F. Roli 49 Methods for fusing multile classifiers Methods for fusing multile classifiers

More information

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation Paer C Exact Volume Balance Versus Exact Mass Balance in Comositional Reservoir Simulation Submitted to Comutational Geosciences, December 2005. Exact Volume Balance Versus Exact Mass Balance in Comositional

More information

Estimation of the large covariance matrix with two-step monotone missing data

Estimation of the large covariance matrix with two-step monotone missing data Estimation of the large covariance matrix with two-ste monotone missing data Masashi Hyodo, Nobumichi Shutoh 2, Takashi Seo, and Tatjana Pavlenko 3 Deartment of Mathematical Information Science, Tokyo

More information

Computer arithmetic. Intensive Computation. Annalisa Massini 2017/2018

Computer arithmetic. Intensive Computation. Annalisa Massini 2017/2018 Comuter arithmetic Intensive Comutation Annalisa Massini 7/8 Intensive Comutation - 7/8 References Comuter Architecture - A Quantitative Aroach Hennessy Patterson Aendix J Intensive Comutation - 7/8 3

More information

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests

System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests 009 American Control Conference Hyatt Regency Riverfront, St. Louis, MO, USA June 0-, 009 FrB4. System Reliability Estimation and Confidence Regions from Subsystem and Full System Tests James C. Sall Abstract

More information

arxiv:cond-mat/ v1 [cond-mat.stat-mech] 27 Jul 2005

arxiv:cond-mat/ v1 [cond-mat.stat-mech] 27 Jul 2005 Self-organized Boolean game on networs arxiv:cond-mat/050766v1 [cond-mat.stat-mech] 7 Jul 005 Tao Zhou 1,, Bing-Hong Wang 1, Pei-Ling Zhou, Chun-Xia Yang, and Jun Liu 1 Deartment of Modern Physics, University

More information

Comparative study on different walking load models

Comparative study on different walking load models Comarative study on different walking load models *Jining Wang 1) and Jun Chen ) 1), ) Deartment of Structural Engineering, Tongji University, Shanghai, China 1) 1510157@tongji.edu.cn ABSTRACT Since the

More information

Detection Algorithm of Particle Contamination in Reticle Images with Continuous Wavelet Transform

Detection Algorithm of Particle Contamination in Reticle Images with Continuous Wavelet Transform Detection Algorithm of Particle Contamination in Reticle Images with Continuous Wavelet Transform Chaoquan Chen and Guoing Qiu School of Comuter Science and IT Jubilee Camus, University of Nottingham Nottingham

More information

Decomposition and Metaoptimization of Mutation Operator in Differential Evolution

Decomposition and Metaoptimization of Mutation Operator in Differential Evolution Decomposition and Metaoptimization of Mutation Operator in Differential Evolution Karol Opara 1 and Jaros law Arabas 2 1 Systems Research Institute, Polish Academy of Sciences 2 Institute of Electronic

More information

VIBRATION ANALYSIS OF BEAMS WITH MULTIPLE CONSTRAINED LAYER DAMPING PATCHES

VIBRATION ANALYSIS OF BEAMS WITH MULTIPLE CONSTRAINED LAYER DAMPING PATCHES Journal of Sound and Vibration (998) 22(5), 78 85 VIBRATION ANALYSIS OF BEAMS WITH MULTIPLE CONSTRAINED LAYER DAMPING PATCHES Acoustics and Dynamics Laboratory, Deartment of Mechanical Engineering, The

More information

Topology Optimization of Three Dimensional Structures under Self-weight and Inertial Forces

Topology Optimization of Three Dimensional Structures under Self-weight and Inertial Forces 6 th World Congresses of Structural and Multidiscilinary Otimization Rio de Janeiro, 30 May - 03 June 2005, Brazil Toology Otimization of Three Dimensional Structures under Self-weight and Inertial Forces

More information

ESTIMATION OF THE RECIPROCAL OF THE MEAN OF THE INVERSE GAUSSIAN DISTRIBUTION WITH PRIOR INFORMATION

ESTIMATION OF THE RECIPROCAL OF THE MEAN OF THE INVERSE GAUSSIAN DISTRIBUTION WITH PRIOR INFORMATION STATISTICA, anno LXVIII, n., 008 ESTIMATION OF THE RECIPROCAL OF THE MEAN OF THE INVERSE GAUSSIAN DISTRIBUTION WITH PRIOR INFORMATION 1. INTRODUCTION The Inverse Gaussian distribution was first introduced

More information

The Value of Even Distribution for Temporal Resource Partitions

The Value of Even Distribution for Temporal Resource Partitions The Value of Even Distribution for Temoral Resource Partitions Yu Li, Albert M. K. Cheng Deartment of Comuter Science University of Houston Houston, TX, 7704, USA htt://www.cs.uh.edu Technical Reort Number

More information

A New Method of DDB Logical Structure Synthesis Using Distributed Tabu Search

A New Method of DDB Logical Structure Synthesis Using Distributed Tabu Search A New Method of DDB Logical Structure Synthesis Using Distributed Tabu Search Eduard Babkin and Margarita Karunina 2, National Research University Higher School of Economics Det of nformation Systems and

More information

PROBABILITY OF FAILURE OF MONOPILE FOUNDATIONS BASED ON LABORATORY MEASUREMENTS

PROBABILITY OF FAILURE OF MONOPILE FOUNDATIONS BASED ON LABORATORY MEASUREMENTS Proceedings of the 6 th International Conference on the Alication of Physical Modelling in Coastal and Port Engineering and Science (Coastlab16) Ottawa, Canada, May 10-13, 2016 Coyright : Creative Commons

More information

The science of making more torque from wind: Diffuser experiments and theory revisited.

The science of making more torque from wind: Diffuser experiments and theory revisited. Journal of Physics: Conference Series The science of making more torque from wind: Diffuser exeriments and theory revisited. To cite this article: Dr Gerard J W van Bussel 7 J. Phys.: Conf. Ser. 75 View

More information

ON OPTIMIZATION OF THE MEASUREMENT MATRIX FOR COMPRESSIVE SENSING

ON OPTIMIZATION OF THE MEASUREMENT MATRIX FOR COMPRESSIVE SENSING 8th Euroean Signal Processing Conference (EUSIPCO-2) Aalborg, Denmark, August 23-27, 2 ON OPTIMIZATION OF THE MEASUREMENT MATRIX FOR COMPRESSIVE SENSING Vahid Abolghasemi, Saideh Ferdowsi, Bahador Makkiabadi,2,

More information

s v 0 q 0 v 1 q 1 v 2 (q 2) v 3 q 3 v 4

s v 0 q 0 v 1 q 1 v 2 (q 2) v 3 q 3 v 4 Discrete Adative Transmission for Fading Channels Lang Lin Λ, Roy D. Yates, Predrag Sasojevic WINLAB, Rutgers University 7 Brett Rd., NJ- fllin, ryates, sasojevg@winlab.rutgers.edu Abstract In this work

More information

#A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS

#A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS #A64 INTEGERS 18 (2018) APPLYING MODULAR ARITHMETIC TO DIOPHANTINE EQUATIONS Ramy F. Taki ElDin Physics and Engineering Mathematics Deartment, Faculty of Engineering, Ain Shams University, Cairo, Egyt

More information

Optimal Design of Truss Structures Using a Neutrosophic Number Optimization Model under an Indeterminate Environment

Optimal Design of Truss Structures Using a Neutrosophic Number Optimization Model under an Indeterminate Environment Neutrosohic Sets and Systems Vol 14 016 93 University of New Mexico Otimal Design of Truss Structures Using a Neutrosohic Number Otimization Model under an Indeterminate Environment Wenzhong Jiang & Jun

More information

2-D Analysis for Iterative Learning Controller for Discrete-Time Systems With Variable Initial Conditions Yong FANG 1, and Tommy W. S.

2-D Analysis for Iterative Learning Controller for Discrete-Time Systems With Variable Initial Conditions Yong FANG 1, and Tommy W. S. -D Analysis for Iterative Learning Controller for Discrete-ime Systems With Variable Initial Conditions Yong FANG, and ommy W. S. Chow Abstract In this aer, an iterative learning controller alying to linear

More information

Approximating min-max k-clustering

Approximating min-max k-clustering Aroximating min-max k-clustering Asaf Levin July 24, 2007 Abstract We consider the roblems of set artitioning into k clusters with minimum total cost and minimum of the maximum cost of a cluster. The cost

More information

Adaptive estimation with change detection for streaming data

Adaptive estimation with change detection for streaming data Adative estimation with change detection for streaming data A thesis resented for the degree of Doctor of Philosohy of the University of London and the Diloma of Imerial College by Dean Adam Bodenham Deartment

More information

A Game Theoretic Investigation of Selection Methods in Two Population Coevolution

A Game Theoretic Investigation of Selection Methods in Two Population Coevolution A Game Theoretic Investigation of Selection Methods in Two Poulation Coevolution Sevan G. Ficici Division of Engineering and Alied Sciences Harvard University Cambridge, Massachusetts 238 USA sevan@eecs.harvard.edu

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell February 10, 2010 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

SHAPE OPTOMIZATION OF H-BEAM FLANGE FOR MAXIMUM PLASTIC ENERGY DISSIPATION

SHAPE OPTOMIZATION OF H-BEAM FLANGE FOR MAXIMUM PLASTIC ENERGY DISSIPATION The Fourth China-Jaan-Korea Joint Symosium on Otimization of Structural and Mechanical Systems Kunming, Nov. 6-9, 2006, China SHAPE OPTOMIZATION OF H-BEAM FLANGE FOR MAXIMUM PLASTIC ENERGY DISSIPATION

More information

Research Article An iterative Algorithm for Hemicontractive Mappings in Banach Spaces

Research Article An iterative Algorithm for Hemicontractive Mappings in Banach Spaces Abstract and Alied Analysis Volume 2012, Article ID 264103, 11 ages doi:10.1155/2012/264103 Research Article An iterative Algorithm for Hemicontractive Maings in Banach Saces Youli Yu, 1 Zhitao Wu, 2 and

More information

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels Aroximate Dynamic Programming for Dynamic Caacity Allocation with Multile Priority Levels Alexander Erdelyi School of Oerations Research and Information Engineering, Cornell University, Ithaca, NY 14853,

More information

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System

Robust Predictive Control of Input Constraints and Interference Suppression for Semi-Trailer System Vol.7, No.7 (4),.37-38 htt://dx.doi.org/.457/ica.4.7.7.3 Robust Predictive Control of Inut Constraints and Interference Suression for Semi-Trailer System Zhao, Yang Electronic and Information Technology

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

Preconditioning techniques for Newton s method for the incompressible Navier Stokes equations

Preconditioning techniques for Newton s method for the incompressible Navier Stokes equations Preconditioning techniques for Newton s method for the incomressible Navier Stokes equations H. C. ELMAN 1, D. LOGHIN 2 and A. J. WATHEN 3 1 Deartment of Comuter Science, University of Maryland, College

More information

On split sample and randomized confidence intervals for binomial proportions

On split sample and randomized confidence intervals for binomial proportions On slit samle and randomized confidence intervals for binomial roortions Måns Thulin Deartment of Mathematics, Usala University arxiv:1402.6536v1 [stat.me] 26 Feb 2014 Abstract Slit samle methods have

More information

A New Quantum Tunneling Particle Swarm Optimization Algorithm for Training Feedforward Neural Networks

A New Quantum Tunneling Particle Swarm Optimization Algorithm for Training Feedforward Neural Networks I.J. Intelligent Systems and Alications, 2018, 11, 64-75 Published Online November 2018 in MECS (htt://www.mecs-ress.org/) DOI: 10.5815/ijisa.2018.11.07 A New Quantum Tunneling Particle Swarm Otimization

More information

Probability Estimates for Multi-class Classification by Pairwise Coupling

Probability Estimates for Multi-class Classification by Pairwise Coupling Probability Estimates for Multi-class Classification by Pairwise Couling Ting-Fan Wu Chih-Jen Lin Deartment of Comuter Science National Taiwan University Taiei 06, Taiwan Ruby C. Weng Deartment of Statistics

More information

A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST BASED ON THE WEIBULL DISTRIBUTION

A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST BASED ON THE WEIBULL DISTRIBUTION O P E R A T I O N S R E S E A R C H A N D D E C I S I O N S No. 27 DOI:.5277/ord73 Nasrullah KHAN Muhammad ASLAM 2 Kyung-Jun KIM 3 Chi-Hyuck JUN 4 A MIXED CONTROL CHART ADAPTED TO THE TRUNCATED LIFE TEST

More information

Lower Confidence Bound for Process-Yield Index S pk with Autocorrelated Process Data

Lower Confidence Bound for Process-Yield Index S pk with Autocorrelated Process Data Quality Technology & Quantitative Management Vol. 1, No.,. 51-65, 15 QTQM IAQM 15 Lower onfidence Bound for Process-Yield Index with Autocorrelated Process Data Fu-Kwun Wang * and Yeneneh Tamirat Deartment

More information

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article Available online www.jocr.com Journal of Chemical and harmaceutical Research, 04, 6(5):904-909 Research Article ISSN : 0975-7384 CODEN(USA) : JCRC5 Robot soccer match location rediction and the alied research

More information

Pairwise active appearance model and its application to echocardiography tracking

Pairwise active appearance model and its application to echocardiography tracking Pairwise active aearance model and its alication to echocardiograhy tracking S. Kevin Zhou 1, J. Shao 2, B. Georgescu 1, and D. Comaniciu 1 1 Integrated Data Systems, Siemens Cororate Research, Inc., Princeton,

More information

An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators

An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators S. K. Mallik, Student Member, IEEE, S. Chakrabarti, Senior Member, IEEE, S. N. Singh, Senior Member, IEEE Deartment of Electrical

More information

The Recursive Fitting of Multivariate. Complex Subset ARX Models

The Recursive Fitting of Multivariate. Complex Subset ARX Models lied Mathematical Sciences, Vol. 1, 2007, no. 23, 1129-1143 The Recursive Fitting of Multivariate Comlex Subset RX Models Jack Penm School of Finance and lied Statistics NU College of Business & conomics

More information

Convergence performance of the coupled-wave and the differential methods for thin gratings

Convergence performance of the coupled-wave and the differential methods for thin gratings Convergence erformance of the couled-wave and the differential methods for thin gratings Philie Lalanne To cite this version: Philie Lalanne. Convergence erformance of the couled-wave and the differential

More information

A Parallel Algorithm for Minimization of Finite Automata

A Parallel Algorithm for Minimization of Finite Automata A Parallel Algorithm for Minimization of Finite Automata B. Ravikumar X. Xiong Deartment of Comuter Science University of Rhode Island Kingston, RI 02881 E-mail: fravi,xiongg@cs.uri.edu Abstract In this

More information

Yixi Shi. Jose Blanchet. IEOR Department Columbia University New York, NY 10027, USA. IEOR Department Columbia University New York, NY 10027, USA

Yixi Shi. Jose Blanchet. IEOR Department Columbia University New York, NY 10027, USA. IEOR Department Columbia University New York, NY 10027, USA Proceedings of the 2011 Winter Simulation Conference S. Jain, R. R. Creasey, J. Himmelsach, K. P. White, and M. Fu, eds. EFFICIENT RARE EVENT SIMULATION FOR HEAVY-TAILED SYSTEMS VIA CROSS ENTROPY Jose

More information

Covariance Matrix Estimation for Reinforcement Learning

Covariance Matrix Estimation for Reinforcement Learning Covariance Matrix Estimation for Reinforcement Learning Tomer Lancewicki Deartment of Electrical Engineering and Comuter Science University of Tennessee Knoxville, TN 37996 tlancewi@utk.edu Itamar Arel

More information

A Unified 2D Representation of Fuzzy Reasoning, CBR, and Experience Based Reasoning

A Unified 2D Representation of Fuzzy Reasoning, CBR, and Experience Based Reasoning University of Wollongong Research Online Faculty of Commerce - aers (Archive) Faculty of Business 26 A Unified 2D Reresentation of Fuzzy Reasoning, CBR, and Exerience Based Reasoning Zhaohao Sun University

More information

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition TNN-2007-P-0332.R1 1 Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Haiing Lu, K.N. Plataniotis and A.N. Venetsanooulos The Edward S. Rogers

More information

Minimax Design of Nonnegative Finite Impulse Response Filters

Minimax Design of Nonnegative Finite Impulse Response Filters Minimax Design of Nonnegative Finite Imulse Resonse Filters Xiaoing Lai, Anke Xue Institute of Information and Control Hangzhou Dianzi University Hangzhou, 3118 China e-mail: laix@hdu.edu.cn; akxue@hdu.edu.cn

More information

Proceedings of the 2012 Winter Simulation Conference C. Laroque, J. Himmelspach, R. Pasupathy, O. Rose, and A.M. Uhrmacher, eds

Proceedings of the 2012 Winter Simulation Conference C. Laroque, J. Himmelspach, R. Pasupathy, O. Rose, and A.M. Uhrmacher, eds Proceedings of the 01 Winter Simulation Conference C. Laroque, J. Himmelsach, R. Pasuathy, O. Rose, and A.M. Uhrmacher, eds OPTIMIZATION VIA GRADIENT ORIENTED POLAR RANDOM SEARCH Haobin Li Loo Hay Lee

More information

Distributed Rule-Based Inference in the Presence of Redundant Information

Distributed Rule-Based Inference in the Presence of Redundant Information istribution Statement : roved for ublic release; distribution is unlimited. istributed Rule-ased Inference in the Presence of Redundant Information June 8, 004 William J. Farrell III Lockheed Martin dvanced

More information

MODULAR LINEAR TRANSVERSE FLUX RELUCTANCE MOTORS

MODULAR LINEAR TRANSVERSE FLUX RELUCTANCE MOTORS MODULAR LINEAR TRANSVERSE FLUX RELUCTANCE MOTORS Dan-Cristian POPA, Vasile IANCU, Loránd SZABÓ, Deartment of Electrical Machines, Technical University of Cluj-Naoca RO-400020 Cluj-Naoca, Romania; e-mail:

More information

Recursive Estimation of the Preisach Density function for a Smart Actuator

Recursive Estimation of the Preisach Density function for a Smart Actuator Recursive Estimation of the Preisach Density function for a Smart Actuator Ram V. Iyer Deartment of Mathematics and Statistics, Texas Tech University, Lubbock, TX 7949-142. ABSTRACT The Preisach oerator

More information

A Comparison between Biased and Unbiased Estimators in Ordinary Least Squares Regression

A Comparison between Biased and Unbiased Estimators in Ordinary Least Squares Regression Journal of Modern Alied Statistical Methods Volume Issue Article 7 --03 A Comarison between Biased and Unbiased Estimators in Ordinary Least Squares Regression Ghadban Khalaf King Khalid University, Saudi

More information

Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process

Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process Using the Divergence Information Criterion for the Determination of the Order of an Autoregressive Process P. Mantalos a1, K. Mattheou b, A. Karagrigoriou b a.deartment of Statistics University of Lund

More information

Diverse Routing in Networks with Probabilistic Failures

Diverse Routing in Networks with Probabilistic Failures Diverse Routing in Networks with Probabilistic Failures Hyang-Won Lee, Member, IEEE, Eytan Modiano, Senior Member, IEEE, Kayi Lee, Member, IEEE Abstract We develo diverse routing schemes for dealing with

More information

Implementation of a Column Generation Heuristic for Vehicle Scheduling in a Medium-Sized Bus Company

Implementation of a Column Generation Heuristic for Vehicle Scheduling in a Medium-Sized Bus Company 7e Conférence Internationale de MOdélisation et SIMulation - MOSIM 08 - du 31 mars au 2 avril 2008 Paris- France «Modélisation, Otimisation MOSIM 08 et Simulation du 31 mars des Systèmes au 2 avril : 2008

More information

Logistics Optimization Using Hybrid Metaheuristic Approach under Very Realistic Conditions

Logistics Optimization Using Hybrid Metaheuristic Approach under Very Realistic Conditions 17 th Euroean Symosium on Comuter Aided Process Engineering ESCAPE17 V. Plesu and P.S. Agachi (Editors) 2007 Elsevier B.V. All rights reserved. 1 Logistics Otimization Using Hybrid Metaheuristic Aroach

More information

A Bound on the Error of Cross Validation Using the Approximation and Estimation Rates, with Consequences for the Training-Test Split

A Bound on the Error of Cross Validation Using the Approximation and Estimation Rates, with Consequences for the Training-Test Split A Bound on the Error of Cross Validation Using the Aroximation and Estimation Rates, with Consequences for the Training-Test Slit Michael Kearns AT&T Bell Laboratories Murray Hill, NJ 7974 mkearns@research.att.com

More information

Crossover and the Different Faces of Differential Evolution Searches

Crossover and the Different Faces of Differential Evolution Searches WCCI 21 IEEE World Congress on Computational Intelligence July, 18-23, 21 - CCIB, Barcelona, Spain CEC IEEE Crossover and the Different Faces of Differential Evolution Searches James Montgomery Abstract

More information

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition

Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Uncorrelated Multilinear Discriminant Analysis with Regularization and Aggregation for Tensor Object Recognition Haiing Lu, K.N. Plataniotis and A.N. Venetsanooulos The Edward S. Rogers Sr. Deartment of

More information

Performance Assessment of Generalized Differential Evolution 3 with a Given Set of Constrained Multi-Objective Test Problems

Performance Assessment of Generalized Differential Evolution 3 with a Given Set of Constrained Multi-Objective Test Problems Performance Assessment of Generalized Differential Evolution 3 with a Given Set of Constrained Multi-Objective Test Problems Saku Kukkonen, Student Member, IEEE and Jouni Lampinen Abstract This paper presents

More information

Unsupervised Hyperspectral Image Analysis Using Independent Component Analysis (ICA)

Unsupervised Hyperspectral Image Analysis Using Independent Component Analysis (ICA) Unsuervised Hyersectral Image Analysis Using Indeendent Comonent Analysis (ICA) Shao-Shan Chiang Chein-I Chang Irving W. Ginsberg Remote Sensing Signal and Image Processing Laboratory Deartment of Comuter

More information

arxiv: v1 [cs.ro] 24 May 2017

arxiv: v1 [cs.ro] 24 May 2017 A Near-Otimal Searation Princile for Nonlinear Stochastic Systems Arising in Robotic Path Planning and Control Mohammadhussein Rafieisakhaei 1, Suman Chakravorty 2 and P. R. Kumar 1 arxiv:1705.08566v1

More information

Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models

Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models Evaluating Circuit Reliability Under Probabilistic Gate-Level Fault Models Ketan N. Patel, Igor L. Markov and John P. Hayes University of Michigan, Ann Arbor 48109-2122 {knatel,imarkov,jhayes}@eecs.umich.edu

More information

ENHANCING TIMBRE MODEL USING MFCC AND ITS TIME DERIVATIVES FOR MUSIC SIMILARITY ESTIMATION

ENHANCING TIMBRE MODEL USING MFCC AND ITS TIME DERIVATIVES FOR MUSIC SIMILARITY ESTIMATION th Euroean Signal Processing Conference (EUSIPCO ) Bucharest, Romania, August 7-3, ENHANCING TIMBRE MODEL USING AND ITS TIME DERIVATIVES FOR MUSIC SIMILARITY ESTIMATION Franz de Leon, Kirk Martinez Electronics

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell October 25, 2009 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

Network DEA: A Modified Non-radial Approach

Network DEA: A Modified Non-radial Approach Network DEA: A Modified Non-radial Aroach Victor John M. Cantor Deartment of Industrial and Systems Engineering National University of Singaore (NUS), Singaore, Singaore Tel: (+65) 913 40025, Email: victorjohn.cantor@u.nus.edu

More information

Re-entry Protocols for Seismically Active Mines Using Statistical Analysis of Aftershock Sequences

Re-entry Protocols for Seismically Active Mines Using Statistical Analysis of Aftershock Sequences Re-entry Protocols for Seismically Active Mines Using Statistical Analysis of Aftershock Sequences J.A. Vallejos & S.M. McKinnon Queen s University, Kingston, ON, Canada ABSTRACT: Re-entry rotocols are

More information

STA 250: Statistics. Notes 7. Bayesian Approach to Statistics. Book chapters: 7.2

STA 250: Statistics. Notes 7. Bayesian Approach to Statistics. Book chapters: 7.2 STA 25: Statistics Notes 7. Bayesian Aroach to Statistics Book chaters: 7.2 1 From calibrating a rocedure to quantifying uncertainty We saw that the central idea of classical testing is to rovide a rigorous

More information

Sparsity Promoting LMS for Adaptive Feedback Cancellation

Sparsity Promoting LMS for Adaptive Feedback Cancellation 7 5th Euroean Signal Processing Conference (EUSIPCO) Sarsity Promoting MS for Adative Feedback Cancellation Ching-Hua ee, Bhaskar D. Rao, and Harinath Garudadri Deartment of Electrical and Comuter Engineering

More information

Meshless Methods for Scientific Computing Final Project

Meshless Methods for Scientific Computing Final Project Meshless Methods for Scientific Comuting Final Project D0051008 洪啟耀 Introduction Floating island becomes an imortant study in recent years, because the lands we can use are limit, so eole start thinking

More information

The non-stochastic multi-armed bandit problem

The non-stochastic multi-armed bandit problem Submitted for journal ublication. The non-stochastic multi-armed bandit roblem Peter Auer Institute for Theoretical Comuter Science Graz University of Technology A-8010 Graz (Austria) auer@igi.tu-graz.ac.at

More information

On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm

On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm On Line Parameter Estimation of Electric Systems using the Bacterial Foraging Algorithm Gabriel Noriega, José Restreo, Víctor Guzmán, Maribel Giménez and José Aller Universidad Simón Bolívar Valle de Sartenejas,

More information

Neural network models for river flow forecasting

Neural network models for river flow forecasting Neural network models for river flow forecasting Nguyen Tan Danh Huynh Ngoc Phien* and Ashim Das Guta Asian Institute of Technology PO Box 4 KlongLuang 220 Pathumthani Thailand Abstract In this study back

More information

On the Relationship Between Packet Size and Router Performance for Heavy-Tailed Traffic 1

On the Relationship Between Packet Size and Router Performance for Heavy-Tailed Traffic 1 On the Relationshi Between Packet Size and Router Performance for Heavy-Tailed Traffic 1 Imad Antonios antoniosi1@southernct.edu CS Deartment MO117 Southern Connecticut State University 501 Crescent St.

More information