A Self-adaptive Similarity-based Fitness Approximation for Evolutionary Optimization

Size: px
Start display at page:

Download "A Self-adaptive Similarity-based Fitness Approximation for Evolutionary Optimization"

Transcription

1 A Self-adaptve Smlarty-based Ftness Approxmaton for Evolutonary Optmzaton Je Tan Dvson of Industral and System Engneerng, Tayuan Unversty of Scence and Technology, Tayuan, 34 Chna College of Informaton Technology, Shandong Womens Unversty, Jnan, 53 Chna Chaol Sun Department of Computer Scence and Technology, Tayuan Unversty of Scence and Technology, Tayuan, 34 Chna State Key Laboratory of Synthetcal Automaton for Process Industres, Northeastern Unversty, Shenyang, 4 chaol.sun.cn@gmal.com Yaochu Jn Department of Computer Scence, Unversty of Surrey, Guldford, GU 7XH, UK Department of Computer Scence and Technology, Tayuan Unversty of Scence and Technology, Tayuan, 34 Chna yaochu.jn@surrey.ac.uk Yn Tan Dvson of Industral and System Engneerng, Tayuan Unversty of Scence and Technology, Tayuan, 34, Chna tanyng965@gmal.com Janchao Zeng School of Computer Scence and Control Engneerng, North Unversty of Chna, Tayuan, 34 Chna zengjanchao@63.net Abstract Evolutonary algorthms used to solve complex optmzaton problems usually need to perform a large number of ftness functon evaluatons, whch often requres huge computatonal overhead. Ths paper proposes a self-adaptve smlarty-based surrogate model as a ftness nhertance strategy to reduce computatonally expensve ftness evaluatons. Gaussan smlarty measurement, whch consders the ruggedness of the landscape, s proposed to adaptvely regulate the smlarty n order to mprove the accuracy of the nhertance ftness values. Emprcal results on three tradtonal benchmark problems wth 5,,, and 3 decson varables and on the CEC 3 test functons wth 3 decson varables demonstrate the hgh effcency and effectveness of the proposed algorthm n that t can obtan better or compettve solutons compared to the state-of-the-art algorthms under a lmted computatonal budget. Keywords Ftness nhertance; Ftness estmaton; smlarty; Computatonally expensve optmzaton; Partcle swarm optmzaton I. INTRODUCTION Evolutonary algorthms (EAs) have much strength for optmzaton, for example, no requrement on dfferentablty of the objectve functons, ease mplementaton, and better global search capablty, compared to tradtonal sngle-pont optmzaton algorthms. For these reasons, EAs have been appled to many real-world applcatons and have emerged as well-establshed and powerful optmzaton tools, especally n the feld of engneerng desgn. However, many real-world engneerng optmzaton problems nvolve complex Ths work was supported n part by Natonal Natural Scence Foundaton of Chna (Grant No , 6437 and 6437), and the State Key Laboratory of Synthetcal Automaton for Process Industres, Northeastern Unversty, Chna. computatonal smulatons, whch may take a few mnutes, hours or even days to perform one smulaton [-4], such as computatonal flud dynamcs [5], fnte element method computatons [6], and aerodynamc shape optmzaton [7], just to name a few. As EAs usually need a large number of ftness evaluatons to obtan an acceptable soluton, t s essental to reduce the number of ftness evaluatons due to the constrant of lmted computatonal resources. To tackle computatonally expensve optmzaton problems, use of computatonally cheap ftness approxmaton to replace the real expensve ftness evaluatons s a common approach [8-]. The most popular method of ftness approxmaton s surrogate models (also called meta-models) [-4]. Commonly used surrogate models nclude the polynomal method [5], the krgng method (Gaussan process) [6], artfcal neural networks (ANN) [7], radal bass functons (RBF) [, 8], and support vector machnes (SVM) [9]. A comprehensve descrpton of these methods can be found n [8-]. Among varous ftness approxmaton methods, ftness nhertance technques are a generc approach that does not rely on the use of surrogate models. One clear beneft of ftness nhertance s that ts computatonal expense s much less than that of tranng a surrogate model [9]. In ths paper, we propose to use a relatvely smple ftness approxmaton method to determne whch ndvdual wll be evaluated usng the real objectve functon, and whch one wll be relably estmated usng ftness nhertance. The dea of ftness nhertance was orgnally proposed by Smth []. It s an ntutve dea assumng that the ftness of an ndvdual can be derved from ts parents. Two types of ftness nhertance strateges have been proposed, one s known as the averaged nhertance and the other weghted nhertance. Salam and Hendtlass [] ntroduced a fast evolutonary algorthm, n

2 whch a ftness value and an assocated relablty value were assgned to each new ndvdual, and the ndvdual s evaluated usng the real ftness functon only when ts relablty value s below a threshold. As mentoned n [], n many real-world applcatons, the ftness of offsprng cannot be relably estmated from the ftness of ther parents, as they are not necessarly smlar to ther parents. Instead, a newly generated offsprng ndvdual may be smlar to those n ts vcnty. For example n a partcle swarm optmzaton (PSO) algorthm, Cu [3] proposed a ftness nhertance model based on the smlarty or credblty to an ndvdual s prevous poston. In [4], a ftness estmaton strategy was proposed for partcle swarm optmzaton, called FESPSO, based on the postonal relatonshp between two ndvduals at same teraton. To further reduce the number of ftness evaluatons for FESPSO, Sun et al [5] ntroduced a smlarty n the ftness estmaton strategy method later. Km [6] used a clusterng technque to group smlar ndvduals and the ndvduals that are closest to cluster center wll be evaluated usng the real objectve functon, whle the ftness of the rest ndvduals wll be nherted from those wth ther ftness beng evaluated usng the real objectve functon. Reyes-Serra and Coello [7] suggested a ftness nhertance model that also used the average method accordng to the Eucldean dstances between ndvduals n multobjectve partcle swarm optmzaton, whch was shown to have reduced the tme of optmzaton and mproved the performance n solvng the classcal test functons. Gomde [8] utlzed a fuzzy clusterng method wth fuzzy adaptve clusterng of c-means. In each cluster, the ftness of the most representatve ndvdual was calculated usng the real ftness functon and the ftness of the other ndvduals s estmated usng a ftness nhertance method weghted by the Eucldean dstance and the membershp degree. Fonseca et al. [9] ntroduced a smlarty-based surrogate model based upon the k-nearest neghbors method (KNN). The smlarty proposed by Fonseca s also based on the Eucldean dstance among ndvduals. Subsequently, Fonseca et al [3] appled three dfferent types of adaptve value nhertance models, ncludng the averaged nhertance model, weghted nhertance model, and parent nhertance model, n a real-coded genetc algorthm for comparsons. Jn and Sendhoff [3] also used the k-means method appled to group the ndvduals of a populaton nto a number of clusters. For each cluster, only the ndvdual whch closest to the cluster center s evaluated and the ftness of other ndvduals are estmated usng a neural network ensemble. The man ssue that needs to be addressed when employng the ftness nhertance method s the accuracy. In most ftness nhertance strateges, the ftness of an ndvdual s estmated based on that of other ndvduals accordng to the postonal relatonshp of these two ndvduals n the decson space usng the Eucldean dstance, and the smaller Eucldean dstance between two ndvduals postons s, the hgher the degree of smlarty between them. However, ftness landscape analyss proved that the dstance metrcs wll become less meanngful when the dmensonalty of the decson space sgnfcantly ncreases [3]. For example, the ftness dstance correlaton (FDC) [33] ndcated that there s no smple relatonshp between the ftness values of two ndvduals and the dstance between two ndvduals. Dfferent from the above-mentoned approaches, Davarynejad [34-36] proposed a method of adaptve fuzzy ftness granulaton usng the ftness of the ndvduals n the pool for ftness estmaton. When the center of the granule has a hgh ftness, the radus of the granules wll be small, and the new ndvdual s ftness s more lkely to be computed n ths granule. On the other hand, more ndvduals wll be estmated usng ftness nhertance n the granule when ts center has a poor ftness. Nonetheless, for fuzzy granule rules, an ndvdual may fnd tself n more than one granule, and granules wth Gaussan measures of smlarty was proposed to adjust ts the wdth so that the new ndvdual wll nhert the ftness from a granule wth a poor ftness yet far from ths ndvdual, whle not nhert the ftness from a granule wth good ftness but near the new ndvdual[37]. Therefore, the above-mentoned approaches cannot guarantee the accuracy n ftness estmaton. In order to obtan more accurate estmaton usng a ftness nhertance method, the mpact of the ftness landscape on the smlarty between two ndvduals must be consdered. In ths paper, an adaptve smlarty-based surrogate model s proposed to reduce the computatonal cost for tmeconsumng optmzaton problems, whch takes nto account the smlarty on both smooth ftness landscapes (wth small ftness changes n the neghborhood) and rough ftness landscapes (wth hgh ftness varatons n the neghborhood). The degree of smlarty between ndvduals wll be decreased when the local ftness landscape s beleved to be rugged. By utlzng an adaptve smlarty-based surrogate models n order to adjust the frequency of usng ftness estmaton, we am to acheve a balance between the frequency and the accuracy of ftness estmaton, thereby ensurng the estmaton accuracy whle reducng the requred ftness evaluatons usng the real ftness functon. The remander of ths paper s organzed as follows. Secton II gves a detaled descrpton of the proposed selfadaptve smlarty-based surrogate model and the surrogateasssted PSO. Expermental results on two groups of test problems are presented and dscussed n Secton III. Secton IV gves a summary of ths paper and some dscussons for future work. II. A SELF-ADAPTIVE SIMILARITY-BASED SURROGATE MODEL As prevously dscussed, the man ssue wth utlzng a ftness nhertance method s to enhance the accuracy of ftness estmaton. In the followng, we frst provde an llustratve example showng why most exstng ftness nhertance methods usng the Eucldean dstance as a measure for smlarty may fal. After that, we ntroduce a new ftness nhertance method. Fg. gves an llustratve example to show the problem n ftness estmaton when a smlarty degree s utlzed n a onedmensonal Mchalewcz functon. In ths example, there are three solutons x, x and x 3, n whch the ftness value of x and x 3 are known and evaluated usng the real ftness functon. x s a new canddate soluton whose ftness s to

3 be estmated. The smlarty between x and x s s, and the smlarty between x 3 and x s s 3. Gven a smlarty threshold s, x wll be estmated f and only f s or s 3 s greater than s. Accordng to the Eucldean dstance, a smaller dstance between two ponts means a hgher degree of smlarty. If s < s < s 3, the ftness of x (y ) can be nherted from that of x 3. However, as we can see from Fg., the ftness landscape s smoother between x and x than between x 3 and x. Therefore, n order to relably estmate the ftness of x, a new smlarty metrc s needed, whch takes the ruggedness of the local landscape n addton to the Eucldean dstance, so that the ftness of x wll be used for estmatng the ftness of x, then the ftness estmaton error wll be smaller than the ftness of x 3 used for estmatng the ftness of x. y X:.5 Y: e-3 x x Mchalewcz,D= X:.5 Y: -.7 x 3 X: Y: x Fg.. An example of three solutons n a one-dmensonal Mchalewcz functon To address the above ssue n ftness estmaton, ths paper proposes a Gaussan smlarty measure that s able to adjust the degree of smlarty accordng to the local ruggedness of the ftness landscape. It s assumed that the smlarty and the local ruggedness are negatvely correlated. A. Self-adaptve smlarty-based ftness approxmaton strategy Wthout loss of generalty, we consder the followng mnmzaton problem: Mnf(x) Subject to : x l x x u () where s the feasble soluton set, n denotes the dmensonalty of the search space. Below are the man steps of our proposed method for PSO algorthms. Step :Intalze a populaton, perform ftness evaluaton usng the real objectve functon for all ndvduals. Step :Determne the personal best poston and global best poston of PSO. If the ftness of the personal best poston or global best poston s estmated, then re-evaluate ther ftness usng real objectve functon. Step 3: Update the velocty and poston of the swarm. For each ndvdual (=,,,n), set a neghborhood sze of x, and select k neghborng ndvduals. If k =, evaluate the ftness of x usng the real objectve functon, and go to Step 6; otherwse go to Step 4. Step 4:Calculate the Gaussan smlarty ¹ r between x and ts neghbor x r. Step 5:Estmate the ftness value of x usng the ftness nhertance strategy. Step 6:Repeat steps -5 f the stop condton s not satsfed, otherwse stop and ext. B. Gaussan Smlarty Gaussan smlarty, whch uses a Gaussan functon, s a smlarty matrx that commonly be used as a clusterng calculaton method for smlarty models, s(; r) = exp( d (x ;x r ) ¾ ) () In (), d (x ; x r ) s the Eucldean dstance between two samples and r, and ¾ s the metrc parameter. The most mportant ssue n the Gaussan smlarty s the selecton of metrcs. Dfferent metrc parameters wll lead to dfferent clusterng results. In ths paper, we ncorporate the roughness of the local ftness landscape to help adjust the metrc of the Gaussan smlarty. The smlarty wll be self-adaptvely calculated accordng to the changes n the local roughness, whch wll reduce the estmaton errors n ftness nhertance. In the followng, we wll elaborate the proposed self-adaptve smlarty-based surrogate model n detal. C. Defnng a measure for local roughness ± It s concevable that the local roughness where ndvdual s located wll nfluence the accuracy of ftness estmaton much more than the global roughness of the ftness landscape. In ths paper, the local roughness wll be consdered when the ftness of ndvdual s to be estmated. Suppose the radus of neghborhood of ndvdual s R, the neghborng U(x ; R ) s defned by [x R ; x + R ]. Apparently, the sze of radus R wll drectly affects the number of ndvduals n the neghborhood. If the neghborhood s too small, there mght not be enough neghborng ndvduals around ndvdual, and the Gaussan smlarty cannot be constructed. If the neghborhood s approprately set, there wll be suffcent neghbors and the ftness of ndvdual wll be nherted from ts neghbors. At the earler generatons, the global search ablty s pad more attenton to the algorthm, so a relatvely large radus R s adopted. Then, R wll be decreased to mprove the accuracy of ftness estmaton at later generatons. In ths paper, an adaptve radus s proposed accordng to the number of the ndvduals dstrbuted n the neghborhood of ndvdual. The upper and lower bounds of the neghborhood on j-th dmenson of ndvdual are defned as cx jmax and cx j mn, that s: cx jmax = max (fx j ; = ; ; ; :::; ng ) (3) cx j mn = mn (fx j ; = ; ; ; :::; ng ) (4) where n s the populaton sze. Correspondngly, the value of R on j-th dmenson s set as: R j = jcx j max cx j mn j (5)

4 In(5), s called the convergence factor. Apparently, R = fr j ; j = ; ; : : : ; lg s dfferent on each dmenson, l s the number of varables. Suppose that there are ndvduals whose ftness s evaluated usng the real objectve functon, we save ther postons and correspondng ftness nto an archve X E = fx ev ; f(x ev ); ev = ;:::; mg, whch wll enable ndvduals to fnd ther neghbors n ther neghborng space U(x ; R ). The neghbors of each ndvdual wll be also saved n an archve X r = fx r ; f(x r ); r = ; :::; kg. x r and f(x r ) represent the poston and ftness of r-th neghbor of ndvdual, respectvely. Eq.(6) defnes the method to calculate the local roughness ± around ndvdual. Obvously, f the ftness varaton among the neghbors s large, the local roughness ± wll be relatvely large, and vce versa. max (f (x r)) mn (f (x r)) rf;;;:::;kg rf;;;:::;kg ± = max (f (x ev)) mn (f (x ev)) (6) evf;;;:::;mg evf;;;:::;mg We can easly fnd that ± [; ]. D. Defnng metrc parameter ¾ The metrc parameter ¾ plays an mportant role n the Gaussan smlarty. We frstly consder the value of the proper metrc parameter ¾ on one-dmensonal problems when the ftness of an ndvdual wll be estmated. The smlarty ¹ r between the nterested ndvdual and ts neghborng ndvduals r s defned accordng to (). ¹ r = e (xr x ) ¾ ; r = ; :::; k (7) The next step s to select those neghbors wth ther smlarty ¹ r hgher than the threshold µ. All neghbors, whch are called estmaton neghbors, wll be used to estmate the ftness of x usng the nhertance method. The values of both parameters, ¾ and µ, are key factors n selectng estmaton neghbors. We dvde ndvduals around ndvdual nto three parts. Accordng to the 3σ prncple of Gaussan functon and the smlarty defned n (7), the poston of x s the center of the neghborhood, and ndvduals n the range [x ¾;x +¾] are called core neghbors, ndvduals located n the range [x ¾; x + ¾] are called outskrt neghbors, and those located n the range [x 3¾; x + 3¾] are called weak outskrt neghbors. When the core neghbors are used to estmate the ftness of ndvdual, the threshold for the smlarty level s set to µ = e ¼ :6. Then the value of ¾ wll drectly affects the range of the core neghbors [x ¾;x +¾]. Therefore, the value of ¾ value wll correspondngly affects the smlarty among ndvduals. In ths paper, a local roughness of the ftness functon ± s proposed to be ncorporated n the settng of metrc parameters, whch ams to adjust the sze of the neghborhood the core neghbors are located. The sze of ths space wll be shrunken as the local roughness ncreases. Obvously, f the roughness s low, the core neghbors wll be selected from a wde range of ndvduals around ndvdual, as a result, the ftness of ndvdual wll be more lkely to be estmated; otherwse, the ftness of ndvdual wll be more lkely to be evaluated usng the real objectve functon. Eq. (8) gves the defnton of ¾ proposed n ths paper: ¾ = ± R (8) where s the convergence factor, and ± represents the local roughness of the ftness functon. Because ± [; ], we can deduce that when the varaton on ftness of neghbors s small n the neghborng space, t s possble for ± to be near zero wth ¾ approachng the nfnty. Therefore, we gve an upper bound of ¾, ¾ R, to set ¾ n a lmted range. Then we can get ¾ = ± R R whch we can derve that ± ; as a result, when ±, we have ± =,and ± [ ; ], and subsequently, ¾ [ R ; R ]. When the ndvduals are n mult-dmensonal space, ¾ and R are n dmensonal vectors wth the same number of dmensons as x, and n the jth dmenson: ¾ j = ± R j (9) The smlarty ¹ r j between ndvdual and ts neghbor r on j-th dmenson s gven n the followng: ¹ r j = e ( xrj x ¾ j j) ; r = ; :::; k () The smlarty ¹ r between x and x r, then, wll be the average of the smlarty on all dmenson: P lj= ¹ r ¹ r j = ; r = ; :::; k () l If the core neghbors are used for estmatng the ftness of x n a weghed manner, the smlarty threshold should be µ = e ¼ :6. We suppose that ndvduals wth a hgher ftness value are commonly surrounded by more ndvduals, so when the average ftness value of the neghborng ndvduals s relatvely hgh, the ndvdual of nterest s more lkely to have a hgher ftness value. Especally at the end of evoluton, the populaton converges gradually, and the ndvdual whose ftness s to be estmated s often located n an area of hgh ftness. As a result, the number k s relatvely hgh n the core neghbors archve X kr = fx kr ; f(x kr ); r = ; :::; k g. Therefore, the ftness of the neghbors should be also consdered when selectng neghbors for ftness estmaton. In ths paper, the threshold of smlarty µ s constructed based on the averaged ftness value of the core neghbors, denoted as mean(f (x kr )), and the threshold µ s adapted accordng to the value of mean(f (x r )): P k kr= mean(f(x kr )) = f(x kr) k ; kr = ; :::; k () mean (f (x kr )) mn (f (x ev)) evf;;;:::;ng µ = :9 :3 max (f (x ev)) mn (f (x ev)) evf;;;:::;ng evf;;;:::;ng (3)

5 From (3) we can see that µ [:6; :9]. By neghborng clusters from the core ndvduals, those neghbors used for ftness estmaton are selected based on ther merts accordng to mean(f (x kr )). When the average ftness mean(f (x kr )) of the ndvduals s weak n the neghborhood, the ftness of the neghbors of x s undesrable overall, and µ wll be reduced accordngly, makng the nterested ndvdual more lkely to be estmated. On the other hand, when the average ftness mean(f (x kr )) of the ndvduals s large n the neghborhood, the value of µ wll be ncreased, thus ncreasng the lkelhood for the ndvdual s ftness to be evaluated usng the real objectve functon. E. Ftness nhertance strategy The ndvduals n the set X r that satsfy ¹ r > µ consttute the set SN = fsn ; :::sn t g. The smlarty between and ndvduals from the set SN s f¹ ; :::¹ t g. Suppose that there are potentally more good ndvduals around the current best ndvdual, there should be the hstorcally best ndvduals found prevously n the set SN, the new ndvdual s also consdered to potentally locate at the best poston. Thus, the ftness of ndvdual s obtaned wth the followng procedure: If the hstorcally best ndvduals are n the set SN, the ftness of ndvdual s evaluated usng the real objectve functon. If t T, P tl= (¹ l ) f(sn l ) f (x ) = P tl= (¹ l ) ; t T (4) If t > T,the smlartes between ndvduals and current ndvdual are ranked accordng to a decreasng order of ther ftness, and T ndvduals were selected to assgn f (x ) value n a weghted manner. T s the maxmum number of neghborng ndvduals used for ftness estmaton. P Tl= (¹ l ) f(sn l ) f (x ) = P Tl= (¹ l ) ; t > T (5) III. EXPERIMENTAL STUDIES In order to examne the performance of the proposed selfadaptve smlarty-based ftness approxmaton for partcle swarm optmzaton, called, for computatonally expensve problems, the algorthm s tested on two sets of wdely used benchmark problems. In addton, the effcency and effectveness of s also demonstrated by comparng to a few exstng ftness nhertance strateges proposed for PSO. A. Tradtonal Optmzaton Benchmark Functons To nvestgate the effcency of our proposed approach, three tradtonal optmzaton benchmark functons suggested n [35] and 8 CEC functons are adopted. Four sets of dmensons are utlzed for three tradtonal optmzaton benchmark functons:d = 5; ; and3. The characterstcs of the three benchmark functons (Grewank, Rastrgn and Ackley), referrng to [35] for more detals. These three benchmark functons are scalable and are commonly used to assess the performance of optmzaton algorthms. They have some ntrgung features whch most optmzaton algorthms fnd hard to deal wth. All the compared algorthms are run for ndependent tmes on each test problem n Matlab R4a. The code for the tradtonal PSO s provded by [38]. In ths paper, The PSO and the k-nearest neghbors method for PSO, called, are adopted to compare on the performance wth under the same fxed ftness evaluatons. In, we select k as the number of the closest neghbors to the ndvduals to be estmated for performng the calculaton of ftness nhertance. The parameters for the method are set as follows: k s the number of the closest neghbors n radus R, whch s the same as the radus set for. From (5) and (6) wth = :5, ftness nhertance s performed for the ndvdual of nterest when À k, but only the closest neghbors to x are selected for estmatng the ftness of ndvdual usng the nhertance strategy when k. The parameters of and are set as follows n the experment: The sze of the swarm s, the cogntve and socal parameters are both set to.5, the velocty v has the same range as the poston x on each dmenson, and the maxmum number of real ftness evaluatons s set to ME = D, whch s the same as those used n [35]. Fgs. -4 shows the results obtaned by PSO,, and wth the same number of ftness evaluatons. The algorthm stops when the ftness evaluatons reach the maxmum value of ftness evaluatons usng the real objectve functon. A number of experments were conducted and showed that f the value of n (5) s set too small, the sze of the neghborhood wll be small too, resultng n an nsuffcent number of the neghborng ndvduals. If the value of n (9) s set too small, the number of the core ndvduals wll be small too, whch wll then greatly reduce the number of ftness estmatons. However, f the value of s too large, there wll be an excessve number of neghborng ndvduals nvolved n ftness evaluatons, whch wll ncrease the tme of ftness approxmaton. In ths paper, ¾ [ R ; R ] was derved when defnng ¾, f the value of s set too large, the range of ¾ s varaton wll be much small, and t wll not be able to properly reflect the mpact of the local roughness on the selecton of neghborng ndvduals used for ftness estmaton. In addton, f the range for selectng core neghbors s too large, t wll lower the accuracy of the ftness estmaton. The values of the parameters n (5) and n (9) are both set by trals and errors. Therefore, n ths experment, = :5, and = :3. Fonseca [9] chose the dfferent largest neghborng number of ndvduals T = ;;5;;5 to verfy the results. The results obtaned when T = ;5 are superor to those obtaned when T = ; ; 5. To mprove the accuracy of ftness nhertance and the computatonal effcency, the maxmum number of neghborng ndvduals nvolved n ftness estmaton T s set to be.

6 Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) Mean (Log) (a) Ackley, (5-D) SS-based FAPSO (c) Ackley, (-D) SS-based FAPSO (b) Ackley, (-D) SS-based FAPSO (d) Ackley, (3-D) SS-based FAPSO Fg.., and SS-based FAPSO Box plot of the best ftness n the fnal generaton for Ackley functon (a) Grewank, (5-D) SS-based FAPSO (c) Grewank, (-D) SS-based FAPSO (b) Grewank, (-D) SS-based FAPSO (d) Grewank, (3-D) SS-based FAPSO Fg. 3., and SS-based FAPSO Boxplots of the best ftness n the fnal generaton for Rastrgn functon (a) Rastrgn, (5-D) SS-based FAPSO (c) Rastrgn, (-D) SS-based FAPSO (b) Rastrgn, (-D) SS-based FAPSO (d) Rastrgn, (3-D) SS-based FAPSO Fg. 4., and SS-based FAPSO Box plot of the best ftness n the fnal generaton for Grewank functon To further demonstrate the performance of ftness nhertance proposed n our method, Fgs. -4 gve the boxplots of the best ftness n the fnal generaton of, and SS-based FAPSO on the three functons. Fg. 5 plots the convergence profle of PSO,, and on benchmark problems (Grewank, Rastrgn and Ackley). As we can see from Fgs. -4, the proposed method has fewer outlers compared to the other two methods n ndependent runs, ndcatng that t performs more robust than the other two algorthms. It can be see that our proposed method obtaned best results nearly n all nstances n terms of the best, the worst and the mean results of the optmal results averaged over ten ndependently runs on three functons of dfferent dmensons compared to PSO and. From Fg. 5, we can see that the proposed SS- Based FAPSO method converges much faster than the other two methods. Although, ntally, the method showed slghtly worse or comparable performance on converge (e.g., n the Rastrgn functon n 3 dmensons), t shows much better convergence capablty as the number of ftness evaluatons ncreases (.e. n the later stages) Ackley, (5-D) Ackley, (-D) Grewank, (5-D) Grewank, (-D) Rastrgn, (5-D) Rastrgn, (-D) Ackley, (-D) Ackley3, (3-D) Grewank, (-D) Grewank 3-D Rastrgn, (-D) Rastrgn3, (3-D) Fg. 5. Convergence profles of the, and on the three benchmark problems B. CEC 3 Functon Tests TABLE I. THE DETAILS OF FUNCTIONS ON WHICH FESPSO AND SS- BASED FAPSO HAVE YIELDED BEST PERFORMANCE. Functon number Detal F Sphere Functon(f) F Rotated Bent Cgar Func Functon(f3) F3 Rotated Rosenbrock s Functon(f6) F4 Rotated Schaffers F7 Functon F5 Composton Functon (n=5,rotated)(f) F6 Composton Functon 8 (n=5,rotated) To further compare the performance of these dfferent methods, the ftness estmaton strategy for partcle swarm

7 optmzaton proposed by Sun et al. (3), called FESPSO, was adopted for comparson. Table II presents the expermental results on CEC 3 testng functons of a dmenson of 3 obtaned by PSO,, and FESPSO, wth a maxmum of, real ftness evaluatons. Table I lsts the detals of the test functons, where we can see that F to F are unmodal functons, F3 to F4 are mult-modal functons, and F5 to F6 are complex combnatons of unmodal and mult-modal functons. TABLE II. COMPARATIVE RESULTS ON THE 8 CEC 3 TEST FUNCTIONS Func Approach Best_ft Worst_ft Ft_mean std_ft opt PSO 7.344E+ 4.34E+3.467E+3.63E+3 F.668E E E E+ -.4 FESPSO 6.776E+.454E E E+3 E E E E E+ F F3 F4 F5 F6 PSO.36E E+.75E E+9.686E+ 4.76E+.54E E+9 FESPSO E+.76E+4.89E E+3.36E+9.747E E E+9 PSO E E+ -5.8E+ 5.48E E E E E+ FESPSO -7.8E+ 7.39E+.796E E E E+ -7.9E E+ PSO E E E E E E E+.7688E+ FESPSO -5.43E E E+3.57E E E E+ 6.3E+ PSO.5673E+3.877E E+3.48E+.546E+3.675E+3.67E E+ FESPSO.9E+3.53E+3.73E+3.766E E+.9E+3.98E E+ PSO.9499E E E+3.746E+ 3.84E E E E+ FESPSO 4.73E E E E+.84E E+3.548E E+ -. E+3-9. E E+ 7. E+.4 E+3 As the results shown n Table II ndcate, method has obtaned better results than the compared algorthms on both unmodal and mult-modal test functons. shows sgnfcantly better overall performance on F. In F4 and F6, the standard devaton obtaned by was slghtly nferor to the compared methods. IV. CONCLUSION AND FUTURE WORK Ths paper proposed a new self-adaptve smlarty-based surrogate model for ftness estmaton n PSO to solvng computatonally expensve problems. In ths model, the radus of the neghborhood to be selected for estmatng the ftness of the ndvduals s self-adaptve. The proposed method requres a lower number of ftness calculatons usng the real ftness functon whle a hgh accuracy n ftness estmatons. Our future work ncludes the development of a more accurate method n evaluatng local roughness of the ftness landscape. Furthermore, the adaptve smlarty-based surrogate model proposed n ths study can be seen as a local surrogate model. Hence, our future work wll consder the stuatons where the condton for ftness nhertance s not satsfed. In these cases, we could combne global surrogate models wth the local estmaton method proposed n ths work to provde more relable ftness estmatons. V. APPENDIX In the canoncal PSO, when searchng n a D-dmensonal hyperspace, each partcle has a velocty V = [v ;v ; :::; v D ] and a poston X = [x ;x ;:::;x D ]. The vectors V and X are ntalzed randomly and then updated by (6) and (7) through the gudance of ts personal best poston P, and the global best poston P n : V (t+) =!V (t) X (t+) + c r (P (t) = X (t) X (t) ) + c r (P (t) n X (t) +V (t+) ) (6) (7) where X (t) and V (t) are the poston and velocty of partcle at generaton t respectvely., The Coeffcents! s called the nerta weght, c and c are acceleraton parameters whch are commonly set to.. r and r are two dagonal matrx whose dagonal elements are random numbers unformly generated wthn the range of [, ]. ACKNOWLEDGMENT Ths work was supported n part by Natonal Natural Scence Foundaton of Chna (Grant No , 6437 and 6437), and the State Key Laboratory of Synthetcal Automaton for Process Industres, Northeastern Unversty, Chna. REFERENCES [] D. Lm, Y. Jn, Y.-S. Ong, and B. Sendhoff, "Generalzng surrogateasssted evolutonary computaton, Evolutonary Computaton, IEEE Transactons on, vol. 4,, pp [] D. Lm, Y.-S. Ong, Y. Jn, and B. Sendhoff, "Trusted evolutonary algorthm, n Evolutonary Computaton, 6. CEC 6. IEEE Congress on, 6, pp [3] Y. S. Ong, P. Nar, A. Keane, and K. Wong, Surrogate-asssted evolutonary optmzaton frameworks for hgh-fdelty engneerng desgn problems, n Knowledge Incorporaton n Evolutonary Computaton, ed: Sprnger, 5, pp [4] Y. S. Ong, P. B. Nar, and A. J. Keane, Evolutonary optmzaton of computatonally expensve problems va surrogate modelng, AIAA journal, vol. 4, 3, pp [5] M. N. Le, Y. S. Ong, S. Menzel, Y. Jn, and B. Sendhoff, Evoluton by adaptng surrogates, Evolutonary computaton, vol., 3, pp [6] K. S. Won and T. Ray, A framework for desgn optmzaton usng surrogates, Engneerng optmzaton, vol. 37, 5, pp [7] M. K. Karakass, A. P. Gots, and K. C. Gannakoglou, Inexact nformaton aded, low cost, dstrbuted genetc algorthms for aerodynamc shape optmzaton, Internatonal Journal for Numercal Methods n Fluds, vol. 43, 3,pp [8] Y. Jn, A comprehensve survey of ftness approxmaton n evolutonary computaton, Soft computng, vol. 9, 5, pp. 3-. [9] L. Sh and K. Rasheed, A survey of ftness approxmaton methods appled n evolutonary algorthms, n Computatonal ntellgence n expensve optmzaton problems, ed: Sprnger,, pp [] Y. Jn, Surrogate-asssted evolutonary computaton: Recent advances and future challenges, Swarm and Evolutonary Computaton, vol.,, pp [] E. Sanchez, S. Pntos, and N. V. Quepo, Toward an optmal ensemble of kernel-based approxmatons wth engneerng applcatons, Structural and multdscplnary optmzaton, vol. 36, 8, pp [] D. Lm, Y.-S. Ong, Y. Jn, and B. Sendhoff, A study on metamodelng technques, ensembles, and mult-surrogates n evolutonary computaton, n Proceedngs of the 9th annual conference on Genetc and evolutonary computaton, 7, pp

8 [3] A. Samad, K.-Y. Km, T. Goel, R. T. Haftka, and W. Shyy, Multple surrogate modelng for axal compressor blade shape optmzaton, Journal of Propulson and Power, vol. 4, 8, pp [4] C. Sun, Y. Jn, J. Zeng, and Y. Yu, A two-layer surrogate-asssted partcle swarm optmzaton algorthm, Soft Computng, vol. 9, 5, pp [5] I. Paenke, J. Branke, and Y. Jn, Effcent search for robust solutons by means of evolutonary algorthms and ftness approxmaton, Evolutonary Computaton, IEEE Transactons on, vol., 6, pp [6] B. Lu, Q. Zhang, and G. G. Gelen, A Gaussan process surrogate model asssted evolutonary algorthm for medum scale expensve optmzaton problems, Evolutonary Computaton, IEEE Transactons on, vol. 8, 4,pp [7] S. Ferrar and R. F. Stengel, Smooth functon approxmaton usng neural networks, Neural Networks, IEEE Transactons on, vol. 6, 5,pp [8] Y. Jn, M. Olhofer, and B. Sendhoff, A framework for evolutonary optmzaton wth approxmate ftness functons, Evolutonary Computaton, IEEE Transactons on, vol. 6,,pp [9] X. Lu, K. Tang, and X. Yao, Classfcaton-asssted dfferental evoluton for computatonally expensve problems, n Evolutonary Computaton (CEC), IEEE Congress on,, pp [] R. E. Smth, B. A. Dke, and S. Stegmann, Ftness nhertance n genetc algorthms, n Proceedngs of the 995 ACM symposum on Appled computng, 995, pp [] M. Salam and T. Hendtlass, A fast evaluaton strategy for evolutonary algorthms, Appled Soft Computng, vol., 3,pp [] E. Ducheyne, B. De Baets, and R. De Wulf, Is ftness nhertance useful for real-world applcatons?, n Evolutonary Mult-Crteron Optmzaton, 3, pp [3] Z. Cu, J. Zeng, and G. Sun, A fast partcle swarm optmzaton, Internatonal Journal of Innovatve Computng, Informaton and Control, vol., 6,pp [4] Z. J. Sun C, Pan J, et al., A new ftness estmaton strategy for partcle swarm optmzaton, Informaton Scences,vol., 3,pp [5] C. Sun, J. Zeng, J. Pan, and Y. Jn, Smlarty-based evoluton control for ftness estmaton n partcle swarm optmzaton, n Computatonal Intellgence n Dynamc and Uncertan Envronments (CIDUE), 3 IEEE Symposum on, 3, pp. -8. [6] H.-S. Km and S.-B. Cho, An effcent genetc algorthm wth less ftness evaluaton by clusterng, n Evolutonary Computaton,. Proceedngs of the Congress on,, pp [7] M. Reyes-Serra and C. A. C. Coello, A study of ftness nhertance and approxmaton technques for mult-objectve partcle swarm optmzaton, n Evolutonary Computaton, 5. The 5 IEEE Congress on, 5, pp [8] F. Gomde, Fuzzy clusterng n ftness estmaton models for genetc algorthms and applcatons, n Fuzzy Systems, 6 IEEE Internatonal Conference on, 6, pp [9] L. Fonseca, H. Barbosa, and A. Lemonge, A smlarty-based surrogate model for enhanced performance n genetc algorthms, Opsearch, vol. 46, 9,pp [3] L. G. Fonseca, A. C. Lemonge, and H. J. Barbosa, A study on ftness nhertance for enhanced effcency n real-coded genetc algorthms, n Evolutonary Computaton (CEC), IEEE Congress on,, pp [3] Y. Jn and B. Sendhoff, Reducng Ftness Evaluatons Usng Clusterng Technques and Neural Network Ensembles, n Genetc and Evolutonary Computaton GECCO 4: Genetc and Evolutonary Computaton Conference, Seattle, WA, USA, June 6-3, 4. Proceedngs, Part I, K. Deb, Ed., ed Berln, Hedelberg: Sprnger Berln Hedelberg, 4, pp [3] Y. Sun, S. K. Halgamuge, M. Krley, and M. A. Munoz, On the selecton of ftness landscape analyss metrcs for contnuous optmzaton problems, n Informaton and Automaton for Sustanablty (ICIAfS), 4 7th Internatonal Conference on, 4, pp. -6. [33] T. Jones and S. Forrest, Ftness Dstance Correlaton as a Measure of Problem Dffculty for Genetc Algorthms, n ICGA, 995, pp [34] M.-R. Akbarzadeh-T, M. Davarynejad, and N. Parz, Adaptve fuzzy ftness granulaton for evolutonary optmzaton, Internatonal Journal of Approxmate Reasonng, vol. 49, 8, pp [35] M. Davarynejad, C. Ahn, J. Vrancken, J. van den Berg, and C. C. Coello, Evolutonary hdden nformaton detecton by granulaton-based ftness approxmaton, Appled Soft Computng, vol.,,pp [36] M. Davarynejad, J. Vrancken, J. van den Berg, and C. A. C. Coello, A ftness granulaton approach for large-scale structural desgn optmzaton, n Varants of Evolutonary Algorthms for Real-World Applcatons, ed: Sprnger,, pp [37] I. Cruz-Vega, M. Garca-Lmon, and H. J. Escalante, Adaptvesurrogate based on a neuro-fuzzy network and granular computng, n Proceedngs of the 4 conference on Genetc and evolutonary computaton, 4, pp [38] J. Lang, B. Qu, P. Suganthan, and A. G. Hernández-Díaz, Problem defntons and evaluaton crtera for the CEC 3 specal sesson on real-parameter optmzaton," Computatonal Intellgence Laboratory, Zhengzhou Unversty, Zhengzhou, Chna and Nanyang Technologcal Unversty, Sngapore, Techncal Report, vol., 3.

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty

More information

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning the Kernel Parameters in Kernel Minimum Distance Classifier Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department

More information

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points; Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features

More information

Multi-objective Optimization Using Self-adaptive Differential Evolution Algorithm

Multi-objective Optimization Using Self-adaptive Differential Evolution Algorithm Mult-objectve Optmzaton Usng Self-adaptve Dfferental Evoluton Algorthm V. L. Huang, S. Z. Zhao, R. Mallpedd and P. N. Suganthan Abstract - In ths paper, we propose a Multobjectve Self-adaptve Dfferental

More information

Cluster Analysis of Electrical Behavior

Cluster Analysis of Electrical Behavior Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School

More information

Meta-heuristics for Multidimensional Knapsack Problems

Meta-heuristics for Multidimensional Knapsack Problems 2012 4th Internatonal Conference on Computer Research and Development IPCSIT vol.39 (2012) (2012) IACSIT Press, Sngapore Meta-heurstcs for Multdmensonal Knapsack Problems Zhbao Man + Computer Scence Department,

More information

Recommended Items Rating Prediction based on RBF Neural Network Optimized by PSO Algorithm

Recommended Items Rating Prediction based on RBF Neural Network Optimized by PSO Algorithm Recommended Items Ratng Predcton based on RBF Neural Network Optmzed by PSO Algorthm Chengfang Tan, Cayn Wang, Yuln L and Xx Q Abstract In order to mtgate the data sparsty and cold-start problems of recommendaton

More information

Determining the Optimal Bandwidth Based on Multi-criterion Fusion

Determining the Optimal Bandwidth Based on Multi-criterion Fusion Proceedngs of 01 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 5 (01) (01) IACSIT Press, Sngapore Determnng the Optmal Bandwdth Based on Mult-crteron Fuson Ha-L Lang 1+, Xan-Mn

More information

CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION

CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION 24 CHAPTER 2 PROPOSED IMPROVED PARTICLE SWARM OPTIMIZATION The present chapter proposes an IPSO approach for multprocessor task schedulng problem wth two classfcatons, namely, statc ndependent tasks and

More information

EVALUATION OF THE PERFORMANCES OF ARTIFICIAL BEE COLONY AND INVASIVE WEED OPTIMIZATION ALGORITHMS ON THE MODIFIED BENCHMARK FUNCTIONS

EVALUATION OF THE PERFORMANCES OF ARTIFICIAL BEE COLONY AND INVASIVE WEED OPTIMIZATION ALGORITHMS ON THE MODIFIED BENCHMARK FUNCTIONS Academc Research Internatonal ISS-L: 3-9553, ISS: 3-9944 Vol., o. 3, May 0 EVALUATIO OF THE PERFORMACES OF ARTIFICIAL BEE COLOY AD IVASIVE WEED OPTIMIZATIO ALGORITHMS O THE MODIFIED BECHMARK FUCTIOS Dlay

More information

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism for Nested Loops with Non-uniform and Flow Dependences Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr

More information

Network Intrusion Detection Based on PSO-SVM

Network Intrusion Detection Based on PSO-SVM TELKOMNIKA Indonesan Journal of Electrcal Engneerng Vol.1, No., February 014, pp. 150 ~ 1508 DOI: http://dx.do.org/10.11591/telkomnka.v1.386 150 Network Intruson Detecton Based on PSO-SVM Changsheng Xang*

More information

Clustering Algorithm Combining CPSO with K-Means Chunqin Gu 1, a, Qian Tao 2, b

Clustering Algorithm Combining CPSO with K-Means Chunqin Gu 1, a, Qian Tao 2, b Internatonal Conference on Advances n Mechancal Engneerng and Industral Informatcs (AMEII 05) Clusterng Algorthm Combnng CPSO wth K-Means Chunqn Gu, a, Qan Tao, b Department of Informaton Scence, Zhongka

More information

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints Australan Journal of Basc and Appled Scences, 2(4): 1204-1208, 2008 ISSN 1991-8178 Sum of Lnear and Fractonal Multobjectve Programmng Problem under Fuzzy Rules Constrants 1 2 Sanjay Jan and Kalash Lachhwan

More information

An Image Fusion Approach Based on Segmentation Region

An Image Fusion Approach Based on Segmentation Region Rong Wang, L-Qun Gao, Shu Yang, Yu-Hua Cha, and Yan-Chun Lu An Image Fuson Approach Based On Segmentaton Regon An Image Fuson Approach Based on Segmentaton Regon Rong Wang, L-Qun Gao, Shu Yang 3, Yu-Hua

More information

The Research of Support Vector Machine in Agricultural Data Classification

The Research of Support Vector Machine in Agricultural Data Classification The Research of Support Vector Machne n Agrcultural Data Classfcaton Le Sh, Qguo Duan, Xnmng Ma, Me Weng College of Informaton and Management Scence, HeNan Agrcultural Unversty, Zhengzhou 45000 Chna Zhengzhou

More information

An Efficient Genetic Algorithm with Fuzzy c-means Clustering for Traveling Salesman Problem

An Efficient Genetic Algorithm with Fuzzy c-means Clustering for Traveling Salesman Problem An Effcent Genetc Algorthm wth Fuzzy c-means Clusterng for Travelng Salesman Problem Jong-Won Yoon and Sung-Bae Cho Dept. of Computer Scence Yonse Unversty Seoul, Korea jwyoon@sclab.yonse.ac.r, sbcho@cs.yonse.ac.r

More information

Support Vector Machines

Support Vector Machines /9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.

More information

Research of Neural Network Classifier Based on FCM and PSO for Breast Cancer Classification

Research of Neural Network Classifier Based on FCM and PSO for Breast Cancer Classification Research of Neural Network Classfer Based on FCM and PSO for Breast Cancer Classfcaton Le Zhang 1, Ln Wang 1, Xujewen Wang 2, Keke Lu 2, and Ajth Abraham 3 1 Shandong Provncal Key Laboratory of Network

More information

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana

More information

Problem Definitions and Evaluation Criteria for the CEC 2015 Competition on Learning-based Real-Parameter Single Objective Optimization

Problem Definitions and Evaluation Criteria for the CEC 2015 Competition on Learning-based Real-Parameter Single Objective Optimization Problem Defntons and Evaluaton Crtera for the CEC 15 Competton on Learnng-based Real-Parameter Sngle Objectve Optmzaton J. J. Lang 1, B. Y. Qu, P. N. Suganthan 3, Q. Chen 4 1 School of Electrcal Engneerng,

More information

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,

More information

S1 Note. Basis functions.

S1 Note. Basis functions. S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type

More information

Smoothing Spline ANOVA for variable screening

Smoothing Spline ANOVA for variable screening Smoothng Splne ANOVA for varable screenng a useful tool for metamodels tranng and mult-objectve optmzaton L. Rcco, E. Rgon, A. Turco Outlne RSM Introducton Possble couplng Test case MOO MOO wth Game Theory

More information

Classifier Selection Based on Data Complexity Measures *

Classifier Selection Based on Data Complexity Measures * Classfer Selecton Based on Data Complexty Measures * Edth Hernández-Reyes, J.A. Carrasco-Ochoa, and J.Fco. Martínez-Trndad Natonal Insttute for Astrophyscs, Optcs and Electroncs, Lus Enrque Erro No.1 Sta.

More information

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal

More information

Wishing you all a Total Quality New Year!

Wishing you all a Total Quality New Year! Total Qualty Management and Sx Sgma Post Graduate Program 214-15 Sesson 4 Vnay Kumar Kalakband Assstant Professor Operatons & Systems Area 1 Wshng you all a Total Qualty New Year! Hope you acheve Sx sgma

More information

Learning-Based Top-N Selection Query Evaluation over Relational Databases

Learning-Based Top-N Selection Query Evaluation over Relational Databases Learnng-Based Top-N Selecton Query Evaluaton over Relatonal Databases Lang Zhu *, Wey Meng ** * School of Mathematcs and Computer Scence, Hebe Unversty, Baodng, Hebe 071002, Chna, zhu@mal.hbu.edu.cn **

More information

Complexity Analysis of Problem-Dimension Using PSO

Complexity Analysis of Problem-Dimension Using PSO Proceedngs of the 7th WSEAS Internatonal Conference on Evolutonary Computng, Cavtat, Croata, June -4, 6 (pp45-5) Complexty Analyss of Problem-Dmenson Usng PSO BUTHAINAH S. AL-KAZEMI AND SAMI J. HABIB,

More information

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth

More information

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique //00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy

More information

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance Tsnghua Unversty at TAC 2009: Summarzng Mult-documents by Informaton Dstance Chong Long, Mnle Huang, Xaoyan Zhu State Key Laboratory of Intellgent Technology and Systems, Tsnghua Natonal Laboratory for

More information

BioTechnology. An Indian Journal FULL PAPER. Trade Science Inc.

BioTechnology. An Indian Journal FULL PAPER. Trade Science Inc. [Type text] [Type text] [Type text] ISSN : 0974-74 Volume 0 Issue BoTechnology 04 An Indan Journal FULL PAPER BTAIJ 0() 04 [684-689] Revew on Chna s sports ndustry fnancng market based on market -orented

More information

Journal of Chemical and Pharmaceutical Research, 2014, 6(6): Research Article. A selective ensemble classification method on microarray data

Journal of Chemical and Pharmaceutical Research, 2014, 6(6): Research Article. A selective ensemble classification method on microarray data Avalable onlne www.ocpr.com Journal of Chemcal and Pharmaceutcal Research, 2014, 6(6):2860-2866 Research Artcle ISSN : 0975-7384 CODEN(USA) : JCPRC5 A selectve ensemble classfcaton method on mcroarray

More information

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning Outlne Artfcal Intellgence and ts applcatons Lecture 8 Unsupervsed Learnng Professor Danel Yeung danyeung@eee.org Dr. Patrck Chan patrckchan@eee.org South Chna Unversty of Technology, Chna Introducton

More information

THE PATH PLANNING ALGORITHM AND SIMULATION FOR MOBILE ROBOT

THE PATH PLANNING ALGORITHM AND SIMULATION FOR MOBILE ROBOT Journal of Theoretcal and Appled Informaton Technology 30 th Aprl 013. Vol. 50 No.3 005-013 JATIT & LLS. All rghts reserved. ISSN: 199-8645 www.jatt.org E-ISSN: 1817-3195 THE PATH PLANNING ALGORITHM AND

More information

Problem Definitions and Evaluation Criteria for the CEC 2005 Special Session on Real-Parameter Optimization

Problem Definitions and Evaluation Criteria for the CEC 2005 Special Session on Real-Parameter Optimization Problem efntons and Evaluaton Crtera for the CEC 2005 Specal Sesson on Real-Parameter Optmzaton P. N. Suganthan, N. Hansen 2, J. J. Lang, K. eb 3, Y. -P. Chen 4, A. Auger 2, S. Twar 3 School of EEE, Nanyang

More information

Maximum Variance Combined with Adaptive Genetic Algorithm for Infrared Image Segmentation

Maximum Variance Combined with Adaptive Genetic Algorithm for Infrared Image Segmentation Internatonal Conference on Logstcs Engneerng, Management and Computer Scence (LEMCS 5) Maxmum Varance Combned wth Adaptve Genetc Algorthm for Infrared Image Segmentaton Huxuan Fu College of Automaton Harbn

More information

Simulation: Solving Dynamic Models ABE 5646 Week 11 Chapter 2, Spring 2010

Simulation: Solving Dynamic Models ABE 5646 Week 11 Chapter 2, Spring 2010 Smulaton: Solvng Dynamc Models ABE 5646 Week Chapter 2, Sprng 200 Week Descrpton Readng Materal Mar 5- Mar 9 Evaluatng [Crop] Models Comparng a model wth data - Graphcal, errors - Measures of agreement

More information

An Optimal Algorithm for Prufer Codes *

An Optimal Algorithm for Prufer Codes * J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,

More information

A Clustering Algorithm Solution to the Collaborative Filtering

A Clustering Algorithm Solution to the Collaborative Filtering Internatonal Journal of Scence Vol.4 No.8 017 ISSN: 1813-4890 A Clusterng Algorthm Soluton to the Collaboratve Flterng Yongl Yang 1, a, Fe Xue, b, Yongquan Ca 1, c Zhenhu Nng 1, d,* Hafeng Lu 3, e 1 Faculty

More information

Feature Reduction and Selection

Feature Reduction and Selection Feature Reducton and Selecton Dr. Shuang LIANG School of Software Engneerng TongJ Unversty Fall, 2012 Today s Topcs Introducton Problems of Dmensonalty Feature Reducton Statstc methods Prncpal Components

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Decson surface s a hyperplane (lne n 2D) n feature space (smlar to the Perceptron) Arguably, the most mportant recent dscovery n machne learnng In a nutshell: map the data to a predetermned

More information

Query Clustering Using a Hybrid Query Similarity Measure

Query Clustering Using a Hybrid Query Similarity Measure Query clusterng usng a hybrd query smlarty measure Fu. L., Goh, D.H., & Foo, S. (2004). WSEAS Transacton on Computers, 3(3), 700-705. Query Clusterng Usng a Hybrd Query Smlarty Measure Ln Fu, Don Hoe-Lan

More information

Review of approximation techniques

Review of approximation techniques CHAPTER 2 Revew of appromaton technques 2. Introducton Optmzaton problems n engneerng desgn are characterzed by the followng assocated features: the objectve functon and constrants are mplct functons evaluated

More information

Edge Detection in Noisy Images Using the Support Vector Machines

Edge Detection in Noisy Images Using the Support Vector Machines Edge Detecton n Nosy Images Usng the Support Vector Machnes Hlaro Gómez-Moreno, Saturnno Maldonado-Bascón, Francsco López-Ferreras Sgnal Theory and Communcatons Department. Unversty of Alcalá Crta. Madrd-Barcelona

More information

Chinese Word Segmentation based on the Improved Particle Swarm Optimization Neural Networks

Chinese Word Segmentation based on the Improved Particle Swarm Optimization Neural Networks Chnese Word Segmentaton based on the Improved Partcle Swarm Optmzaton Neural Networks Ja He Computatonal Intellgence Laboratory School of Computer Scence and Engneerng, UESTC Chengdu, Chna Department of

More information

Machine Learning: Algorithms and Applications

Machine Learning: Algorithms and Applications 14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of

More information

Module Management Tool in Software Development Organizations

Module Management Tool in Software Development Organizations Journal of Computer Scence (5): 8-, 7 ISSN 59-66 7 Scence Publcatons Management Tool n Software Development Organzatons Ahmad A. Al-Rababah and Mohammad A. Al-Rababah Faculty of IT, Al-Ahlyyah Amman Unversty,

More information

GA-Based Learning Algorithms to Identify Fuzzy Rules for Fuzzy Neural Networks

GA-Based Learning Algorithms to Identify Fuzzy Rules for Fuzzy Neural Networks Seventh Internatonal Conference on Intellgent Systems Desgn and Applcatons GA-Based Learnng Algorthms to Identfy Fuzzy Rules for Fuzzy Neural Networks K Almejall, K Dahal, Member IEEE, and A Hossan, Member

More information

NGPM -- A NSGA-II Program in Matlab

NGPM -- A NSGA-II Program in Matlab Verson 1.4 LIN Song Aerospace Structural Dynamcs Research Laboratory College of Astronautcs, Northwestern Polytechncal Unversty, Chna Emal: lsssswc@163.com 2011-07-26 Contents Contents... 1. Introducton...

More information

Solving two-person zero-sum game by Matlab

Solving two-person zero-sum game by Matlab Appled Mechancs and Materals Onlne: 2011-02-02 ISSN: 1662-7482, Vols. 50-51, pp 262-265 do:10.4028/www.scentfc.net/amm.50-51.262 2011 Trans Tech Publcatons, Swtzerland Solvng two-person zero-sum game by

More information

12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification

12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification Introducton to Artfcal Intellgence V22.0472-001 Fall 2009 Lecture 24: Nearest-Neghbors & Support Vector Machnes Rob Fergus Dept of Computer Scence, Courant Insttute, NYU Sldes from Danel Yeung, John DeNero

More information

BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET

BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET 1 BOOSTING CLASSIFICATION ACCURACY WITH SAMPLES CHOSEN FROM A VALIDATION SET TZU-CHENG CHUANG School of Electrcal and Computer Engneerng, Purdue Unversty, West Lafayette, Indana 47907 SAUL B. GELFAND School

More information

ARTICLE IN PRESS. Applied Soft Computing xxx (2012) xxx xxx. Contents lists available at SciVerse ScienceDirect. Applied Soft Computing

ARTICLE IN PRESS. Applied Soft Computing xxx (2012) xxx xxx. Contents lists available at SciVerse ScienceDirect. Applied Soft Computing ASOC-11; o. of Pages 1 Appled Soft Computng xxx (1) xxx xxx Contents lsts avalable at ScVerse ScenceDrect Appled Soft Computng j ourna l ho mepage: www.elsever.com/locate/asoc A herarchcal partcle swarm

More information

PARETO BAYESIAN OPTIMIZATION ALGORITHM FOR THE MULTIOBJECTIVE 0/1 KNAPSACK PROBLEM

PARETO BAYESIAN OPTIMIZATION ALGORITHM FOR THE MULTIOBJECTIVE 0/1 KNAPSACK PROBLEM PARETO BAYESIAN OPTIMIZATION ALGORITHM FOR THE MULTIOBJECTIVE 0/ KNAPSACK PROBLEM Josef Schwarz Jří Očenáše Brno Unversty of Technology Faculty of Engneerng and Computer Scence Department of Computer Scence

More information

BIN XIA et al: AN IMPROVED K-MEANS ALGORITHM BASED ON CLOUD PLATFORM FOR DATA MINING

BIN XIA et al: AN IMPROVED K-MEANS ALGORITHM BASED ON CLOUD PLATFORM FOR DATA MINING An Improved K-means Algorthm based on Cloud Platform for Data Mnng Bn Xa *, Yan Lu 2. School of nformaton and management scence, Henan Agrcultural Unversty, Zhengzhou, Henan 450002, P.R. Chna 2. College

More information

Optimal Design of Nonlinear Fuzzy Model by Means of Independent Fuzzy Scatter Partition

Optimal Design of Nonlinear Fuzzy Model by Means of Independent Fuzzy Scatter Partition Optmal Desgn of onlnear Fuzzy Model by Means of Independent Fuzzy Scatter Partton Keon-Jun Park, Hyung-Kl Kang and Yong-Kab Km *, Department of Informaton and Communcaton Engneerng, Wonkwang Unversty,

More information

Performance Evaluation of Information Retrieval Systems

Performance Evaluation of Information Retrieval Systems Why System Evaluaton? Performance Evaluaton of Informaton Retreval Systems Many sldes n ths secton are adapted from Prof. Joydeep Ghosh (UT ECE) who n turn adapted them from Prof. Dk Lee (Unv. of Scence

More information

Comparison of Heuristics for Scheduling Independent Tasks on Heterogeneous Distributed Environments

Comparison of Heuristics for Scheduling Independent Tasks on Heterogeneous Distributed Environments Comparson of Heurstcs for Schedulng Independent Tasks on Heterogeneous Dstrbuted Envronments Hesam Izakan¹, Ath Abraham², Senor Member, IEEE, Václav Snášel³ ¹ Islamc Azad Unversty, Ramsar Branch, Ramsar,

More information

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 A mathematcal programmng approach to the analyss, desgn and

More information

Adaptive Virtual Support Vector Machine for the Reliability Analysis of High-Dimensional Problems

Adaptive Virtual Support Vector Machine for the Reliability Analysis of High-Dimensional Problems Proceedngs of the ASME 2 Internatonal Desgn Engneerng Techncal Conferences & Computers and Informaton n Engneerng Conference IDETC/CIE 2 August 29-3, 2, Washngton, D.C., USA DETC2-47538 Adaptve Vrtual

More information

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision SLAM Summer School 2006 Practcal 2: SLAM usng Monocular Vson Javer Cvera, Unversty of Zaragoza Andrew J. Davson, Imperal College London J.M.M Montel, Unversty of Zaragoza. josemar@unzar.es, jcvera@unzar.es,

More information

Classifier Swarms for Human Detection in Infrared Imagery

Classifier Swarms for Human Detection in Infrared Imagery Classfer Swarms for Human Detecton n Infrared Imagery Yur Owechko, Swarup Medasan, and Narayan Srnvasa HRL Laboratores, LLC 3011 Malbu Canyon Road, Malbu, CA 90265 {owechko, smedasan, nsrnvasa}@hrl.com

More information

Unsupervised Learning

Unsupervised Learning Pattern Recognton Lecture 8 Outlne Introducton Unsupervsed Learnng Parametrc VS Non-Parametrc Approach Mxture of Denstes Maxmum-Lkelhood Estmates Clusterng Prof. Danel Yeung School of Computer Scence and

More information

Machine Learning 9. week

Machine Learning 9. week Machne Learnng 9. week Mappng Concept Radal Bass Functons (RBF) RBF Networks 1 Mappng It s probably the best scenaro for the classfcaton of two dataset s to separate them lnearly. As you see n the below

More information

Classification / Regression Support Vector Machines

Classification / Regression Support Vector Machines Classfcaton / Regresson Support Vector Machnes Jeff Howbert Introducton to Machne Learnng Wnter 04 Topcs SVM classfers for lnearly separable classes SVM classfers for non-lnearly separable classes SVM

More information

A Similarity Measure Method for Symbolization Time Series

A Similarity Measure Method for Symbolization Time Series Research Journal of Appled Scences, Engneerng and Technology 5(5): 1726-1730, 2013 ISSN: 2040-7459; e-issn: 2040-7467 Maxwell Scentfc Organzaton, 2013 Submtted: July 27, 2012 Accepted: September 03, 2012

More information

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems A Unfed Framework for Semantcs and Feature Based Relevance Feedback n Image Retreval Systems Ye Lu *, Chunhu Hu 2, Xngquan Zhu 3*, HongJang Zhang 2, Qang Yang * School of Computng Scence Smon Fraser Unversty

More information

The Codesign Challenge

The Codesign Challenge ECE 4530 Codesgn Challenge Fall 2007 Hardware/Software Codesgn The Codesgn Challenge Objectves In the codesgn challenge, your task s to accelerate a gven software reference mplementaton as fast as possble.

More information

Available online at Available online at Advanced in Control Engineering and Information Science

Available online at   Available online at   Advanced in Control Engineering and Information Science Avalable onlne at wwwscencedrectcom Avalable onlne at wwwscencedrectcom Proceda Proceda Engneerng Engneerng 00 (2011) 15000 000 (2011) 1642 1646 Proceda Engneerng wwwelsevercom/locate/proceda Advanced

More information

A Robust Method for Estimating the Fundamental Matrix

A Robust Method for Estimating the Fundamental Matrix Proc. VIIth Dgtal Image Computng: Technques and Applcatons, Sun C., Talbot H., Ourseln S. and Adraansen T. (Eds.), 0- Dec. 003, Sydney A Robust Method for Estmatng the Fundamental Matrx C.L. Feng and Y.S.

More information

A Time-driven Data Placement Strategy for a Scientific Workflow Combining Edge Computing and Cloud Computing

A Time-driven Data Placement Strategy for a Scientific Workflow Combining Edge Computing and Cloud Computing > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1 A Tme-drven Data Placement Strategy for a Scentfc Workflow Combnng Edge Computng and Cloud Computng Bng Ln, Fangnng

More information

FINDING IMPORTANT NODES IN SOCIAL NETWORKS BASED ON MODIFIED PAGERANK

FINDING IMPORTANT NODES IN SOCIAL NETWORKS BASED ON MODIFIED PAGERANK FINDING IMPORTANT NODES IN SOCIAL NETWORKS BASED ON MODIFIED PAGERANK L-qng Qu, Yong-quan Lang 2, Jng-Chen 3, 2 College of Informaton Scence and Technology, Shandong Unversty of Scence and Technology,

More information

MULTIOBJECTIVE OPTIMIZATION USING PARALLEL VECTOR EVALUATED PARTICLE SWARM OPTIMIZATION

MULTIOBJECTIVE OPTIMIZATION USING PARALLEL VECTOR EVALUATED PARTICLE SWARM OPTIMIZATION MULTIOBJECTIVE OPTIMIZATION USING PARALLEL VECTOR EVALUATED PARTICLE OPTIMIZATION K.E. Parsopoulos, D.K. Tasouls, M.N. Vrahats Department of Mathematcs, Unversty of Patras Artfcal Intellgence Research

More information

A Simple Methodology for Database Clustering. Hao Tang 12 Guangdong University of Technology, Guangdong, , China

A Simple Methodology for Database Clustering. Hao Tang 12 Guangdong University of Technology, Guangdong, , China for Database Clusterng Guangdong Unversty of Technology, Guangdong, 0503, Chna E-mal: 6085@qq.com Me Zhang Guangdong Unversty of Technology, Guangdong, 0503, Chna E-mal:64605455@qq.com Database clusterng

More information

Fuzzy Rough Neural Network and Its Application to Feature Selection

Fuzzy Rough Neural Network and Its Application to Feature Selection 70 Internatonal Journal of Fuzzy Systems, Vol. 3, No. 4, December 0 Fuzzy Rough Neural Network and Its Applcaton to Feature Selecton Junyang Zhao and Zhl Zhang Abstract For the sake of measurng fuzzy uncertanty

More information

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation Intellgent Informaton Management, 013, 5, 191-195 Publshed Onlne November 013 (http://www.scrp.org/journal/m) http://dx.do.org/10.36/m.013.5601 Qualty Improvement Algorthm for Tetrahedral Mesh Based on

More information

APPLICATION OF MULTIVARIATE LOSS FUNCTION FOR ASSESSMENT OF THE QUALITY OF TECHNOLOGICAL PROCESS MANAGEMENT

APPLICATION OF MULTIVARIATE LOSS FUNCTION FOR ASSESSMENT OF THE QUALITY OF TECHNOLOGICAL PROCESS MANAGEMENT 3. - 5. 5., Brno, Czech Republc, EU APPLICATION OF MULTIVARIATE LOSS FUNCTION FOR ASSESSMENT OF THE QUALITY OF TECHNOLOGICAL PROCESS MANAGEMENT Abstract Josef TOŠENOVSKÝ ) Lenka MONSPORTOVÁ ) Flp TOŠENOVSKÝ

More information

Fast Computation of Shortest Path for Visiting Segments in the Plane

Fast Computation of Shortest Path for Visiting Segments in the Plane Send Orders for Reprnts to reprnts@benthamscence.ae 4 The Open Cybernetcs & Systemcs Journal, 04, 8, 4-9 Open Access Fast Computaton of Shortest Path for Vstng Segments n the Plane Ljuan Wang,, Bo Jang

More information

Face Recognition Method Based on Within-class Clustering SVM

Face Recognition Method Based on Within-class Clustering SVM Face Recognton Method Based on Wthn-class Clusterng SVM Yan Wu, Xao Yao and Yng Xa Department of Computer Scence and Engneerng Tong Unversty Shangha, Chna Abstract - A face recognton method based on Wthn-class

More information

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng

More information

A Binarization Algorithm specialized on Document Images and Photos

A Binarization Algorithm specialized on Document Images and Photos A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a

More information

Design of Structure Optimization with APDL

Design of Structure Optimization with APDL Desgn of Structure Optmzaton wth APDL Yanyun School of Cvl Engneerng and Archtecture, East Chna Jaotong Unversty Nanchang 330013 Chna Abstract In ths paper, the desgn process of structure optmzaton wth

More information

K-means Optimization Clustering Algorithm Based on Hybrid PSO/GA Optimization and CS validity index

K-means Optimization Clustering Algorithm Based on Hybrid PSO/GA Optimization and CS validity index Orgnal Artcle Prnt ISSN: 3-6379 Onlne ISSN: 3-595X DOI: 0.7354/jss/07/33 K-means Optmzaton Clusterng Algorthm Based on Hybrd PSO/GA Optmzaton and CS valdty ndex K Jahanbn *, F Rahmanan, H Rezae 3, Y Farhang

More information

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like: Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A

More information

Positive Semi-definite Programming Localization in Wireless Sensor Networks

Positive Semi-definite Programming Localization in Wireless Sensor Networks Postve Sem-defnte Programmng Localzaton n Wreless Sensor etworks Shengdong Xe 1,, Jn Wang, Aqun Hu 1, Yunl Gu, Jang Xu, 1 School of Informaton Scence and Engneerng, Southeast Unversty, 10096, anjng Computer

More information

Multiple Trajectory Search for Large Scale Global Optimization

Multiple Trajectory Search for Large Scale Global Optimization Multple Trajectory Search for Large Scale Global Optmzaton Ln-YuTsengandChunChen Abstract In ths paper, the multple trajectory search (MTS) s presented for large scale global optmzaton. The MTS uses multple

More information

An Improved Particle Swarm Optimization for Feature Selection

An Improved Particle Swarm Optimization for Feature Selection Journal of Bonc Engneerng 8 (20)?????? An Improved Partcle Swarm Optmzaton for Feature Selecton Yuannng Lu,2, Gang Wang,2, Hulng Chen,2, Hao Dong,2, Xaodong Zhu,2, Sujng Wang,2 Abstract. College of Computer

More information

A Novel Deluge Swarm Algorithm for Optimization Problems

A Novel Deluge Swarm Algorithm for Optimization Problems A Novel eluge Swarm Algorthm for Optmzaton Problems Anahta Samad,* - Mohammad Reza Meybod Scence and Research Branch, Islamc Azad Unversty, Qazvn, Iran Soft Computng Laboratory, Computer Engneerng and

More information

An Evolvable Clustering Based Algorithm to Learn Distance Function for Supervised Environment

An Evolvable Clustering Based Algorithm to Learn Distance Function for Supervised Environment IJCSI Internatonal Journal of Computer Scence Issues, Vol. 7, Issue 5, September 2010 ISSN (Onlne): 1694-0814 www.ijcsi.org 374 An Evolvable Clusterng Based Algorthm to Learn Dstance Functon for Supervsed

More information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

The Shortest Path of Touring Lines given in the Plane

The Shortest Path of Touring Lines given in the Plane Send Orders for Reprnts to reprnts@benthamscence.ae 262 The Open Cybernetcs & Systemcs Journal, 2015, 9, 262-267 The Shortest Path of Tourng Lnes gven n the Plane Open Access Ljuan Wang 1,2, Dandan He

More information

A Simple and Efficient Goal Programming Model for Computing of Fuzzy Linear Regression Parameters with Considering Outliers

A Simple and Efficient Goal Programming Model for Computing of Fuzzy Linear Regression Parameters with Considering Outliers 62626262621 Journal of Uncertan Systems Vol.5, No.1, pp.62-71, 211 Onlne at: www.us.org.u A Smple and Effcent Goal Programmng Model for Computng of Fuzzy Lnear Regresson Parameters wth Consderng Outlers

More information

GSLM Operations Research II Fall 13/14

GSLM Operations Research II Fall 13/14 GSLM 58 Operatons Research II Fall /4 6. Separable Programmng Consder a general NLP mn f(x) s.t. g j (x) b j j =. m. Defnton 6.. The NLP s a separable program f ts objectve functon and all constrants are

More information

High-Boost Mesh Filtering for 3-D Shape Enhancement

High-Boost Mesh Filtering for 3-D Shape Enhancement Hgh-Boost Mesh Flterng for 3-D Shape Enhancement Hrokazu Yagou Λ Alexander Belyaev y Damng We z Λ y z ; ; Shape Modelng Laboratory, Unversty of Azu, Azu-Wakamatsu 965-8580 Japan y Computer Graphcs Group,

More information

SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB

SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB SURFACE PROFILE EVALUATION BY FRACTAL DIMENSION AND STATISTIC TOOLS USING MATLAB V. Hotař, A. Hotař Techncal Unversty of Lberec, Department of Glass Producng Machnes and Robotcs, Department of Materal

More information

Lecture #15 Lecture Notes

Lecture #15 Lecture Notes Lecture #15 Lecture Notes The ocean water column s very much a 3-D spatal entt and we need to represent that structure n an economcal way to deal wth t n calculatons. We wll dscuss one way to do so, emprcal

More information

Virtual Machine Migration based on Trust Measurement of Computer Node

Virtual Machine Migration based on Trust Measurement of Computer Node Appled Mechancs and Materals Onlne: 2014-04-04 ISSN: 1662-7482, Vols. 536-537, pp 678-682 do:10.4028/www.scentfc.net/amm.536-537.678 2014 Trans Tech Publcatons, Swtzerland Vrtual Machne Mgraton based on

More information

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data A Fast Content-Based Multmeda Retreval Technque Usng Compressed Data Borko Furht and Pornvt Saksobhavvat NSF Multmeda Laboratory Florda Atlantc Unversty, Boca Raton, Florda 3343 ABSTRACT In ths paper,

More information