Case Mining from Large Databases

Size: px
Start display at page:

Download "Case Mining from Large Databases"

Transcription

1 Case Mnng from Large Databases Qang Yang and Hong Cheng Department of Computer Scence, Hong Kong Unversty of Scence and Technology, Clearwater Bay, Kowloon Hong Kong {qyang, Abstract. Ths paper presents an approach of case mnng to automatcally dscover case bases from large datasets n order to mprove both the speed and the qualty of case based reasonng. Case mnng constructs a case base from a large raw dataset wth an objectve to mprove the case-base reasonng systems effcency and qualty. Our approach starts from a raw database of objects wth class attrbutes together wth a hstorcal database of past acton sequences on these objects. The object databases can be customer records and the hstorcal acton logs can be the techncal advses gven to the customers to solve ther problems. Our goal s to dscover effectve and hghly representatve problem descrptons assocated wth soluton plans that accomplsh ther tasks. To mantan effcency of computaton, data mnng methods are employed n the process of composng the case base. We motvate the applcaton of the case mnng model usng a fnancal applcaton example, and demonstrate the effectveness of the model usng both real and smulated datasets. 1 Introducton To motvate the case mnng problem, consder a fnancal applcaton example. Suppose that a certan ABC Bank s nterested n ntatng a marketng campagn to encourage ts customers to sgn up for a new loan program. Suppose that we are gven two datasets on (1) customer nformaton and (2) past marketng acton logs on customers n (1) and the fnal outcomes n loan-sgnup results. Tables 1 and 2 show an example of a customer database table together wth some examples of past marketng actons on customers. Suppose that we are nterested n buldng a campagn plan for a new customer Steve. Based on the past campagn actons for people lke Steve, there are many canddate actons that one can suggeston. For example, we can use two of the past cases for John and Mary. If we follow the plan for John, we can frst gve hm a personal phone call and then send hm a gft. Alternatvely, we can decrease the mortgage rate for Steve, followed by offerng a new Credt card. The latter plan follows that of Mary s success. In ether stuaton, the ABC bank mght have a relatvely hgh chance of convertng Steve from a reluctant customer to a wllng one. The above problem can be formulated as a case-based reasonng problem, where the key ssue s to fnd a good role model for customers such as Steve, and apply case 1

2 retreval and adaptaton technques [9,10] to fnd low-cost plans for these customers on a case-by-case bass. Our approach s to frst dentfy typcal problem descrptons as potental role models from the customer database (Table 1). Then for each typcal problem, a cost-effectve plan s found to be assocated wth each negatve-class role model (Table 2); they are stored n a case-base as a plan. Then, for each new problem, we fnd a smlar problem descrpton and adapt ts plan n the case base. Table 1. An example of customer database. The last attrbute s the class attrbute. Customer Salary Cars Mortgage Loan John 80K 3 None Y Mary 40K 1 300K N Steve 40K 1 None N Table 2. Acton-log database example contanng marketng actons and outcomes Advsee Acton1 State1 Acton2 State2 John Phone Featurevalues Send Feature- call for gft values for John John Mary Lower Featurevalues Credt Featurevalues mortgage for Mary card Mary for In ths paper, we focus on how to fnd the case base from a gven database effcently. A challengng ssue facng the constructon of the case base s the large sze of the database and large amount of data. To solve the scale-up problem, we explot data mnng technques, ncludng clusterng and assocaton rule mnng algorthms [2, 5, 8, 14]. Cluster center objects, also known as the medods, are obtaned from the customer database to produce the bass for cases from customers belongng to both the negatve class (problem descrptons for new problems) and the postve class (where Class=Yes n the fnal results table). Assocaton-rule mnng s appled to the actonlog database to produce typcal plans to be assocated wth the problem-objectve pars. The resultant case base s then used for problem solvng for new customers. Because we extract hgh-qualty cases from large databases usng data mnng technques, we call ths method case mnng. Emprcal results are obtaned to demonstrate that the case-mnng algorthms are scalable and acheve a balance between qualty and effcency. 2

3 2 Related Work Case mnng s closely related to case-base mantenance, whereby the case base structure, ndex and knowledge representaton are updated to mprove problemsolvng performance. The most recent specal ssue of Computatonal Intellgence Journal on case-based mantenance [6] hghlghted ths nterest and progress. Smyth and Keane [15], who defned case-base competences based on how cases n an exstng case base are related to each other, addressed the mantenance ssue by proposng a deleton-based algorthm. Ther mantenance polcy s amed at removng as much redundancy form a case base as possble by reducng the overlap between cases or groups of cases. Subsequently, Smyth and McKenna [22] and Yang and Zhu [16] explored how other polces can be appled to case-based mantenance. Leake, Kngly and Wlson [10, 11, 17] analyzed the case-base mantenance problems and cast them nto a general framework of revsng the content and structure of an exstng case base. Aamodt et. al [1] presented learnng algorthms for smlarty functons used n case retreval. In our formulaton, case mnng s targeted for large, un-ndexed raw databases rather than an exstng case base. The ndexng problems have been an ssue of ntensve study n the past, ncludng work n [4, 19, 20]. The ndexng lterature assumes that there s already a case base n exstence, whereas n ths paper we explore how to fnd the case base from hstorcal databases. Patterson et. Al [21] dscussed how to apply the K-means algorthm for case-base mantenance. The work s also related to case-based plannng [9, 18], whch derves a new plan based on plans stored n case bases. However, case-base plannng has manly focused on case adaptaton; nstead, we focus on how to explot the dstrbuton of massve data to obtan statstcally well-balanced case bases. 3 Mnng Problem Descrptons Our algorthm s dvded nto two phases. In phase one, we take the customer database and fnd a set of typcal cases by applyng clusterng algorthm to t. The result s a set of postve-class and negatve-class objects. In order to fnd the typcal cases, we apply two dfferent algorthms ncludng a clusterng algorthm and an SVM-based algorthm. In phase two, we fnd a plan for each negatve-class role model to form a potental case. Care s taken n the search process to fnd only hgh-utlty plans. We wsh to fnd K typcal cases to populate the case base, where the parameter K s user defned; later we dscuss another method (SVM) whch does not requre K to be specfed. These cases should be hghly representatve of the dstrbuton of the customer nformaton n the orgnal customer database. They wll serve as the ntal states for our planner to work on n the next phase. We consder all negatve nstances n the database as the tranng data for mnng the problem descrptons of the case base. Our frst method apples the clusterng algorthms to the negatve nstances, as descrbed n more detal n Table 3. The tranng database conssts of the negatve nstances of the orgnal database. 3

4 Table 3. Algorthm Centrods-CBMne (database DB, nt K) Steps Begn 1 casebase= emptyset; 2 DB = RemoveIrrelevantAttrbutes(DB); 3 Separate the DB nto DB+ and DB-; 4 Clusters+ = ApplyKMeans(DB-, K); 5 for each cluster n Clusters+, do 6 C = fndcentrod(cluster); 7 Insert(C, casebase); 8 end for; 9 Return casebase; End In the algorthm Centrods-CBMne n Table 3, the nput database DB s a raw database. There are two classes n ths database, where the negatve class corresponds to populaton of ntal states for a marketng plan. Step 2 of the algorthm performs feature extracton by applyng a feature flter to the database to remove all attrbutes that are consdered low n nformaton content. For example, f two attrbutes A1 and A2 n the database are hghly correlated, then one of them can be removed. Smlarly, f an attrbute A has very low classfcaton power for the data as can be computed usng nformaton theory, then t can be removed as well. In our mplementaton, we apply a C4.5 decson-tree learnng algorthm to the database DB. After a decson tree s constructed, the attrbutes that are not contaned n the tree are removed from the database; these are the rrelevant attrbutes. Step 3 of the algorthm separates the tranng database nto two parttons, a postve-class subset and a negatve-class subset. Step 4 of the algorthm performs the K- means clusterng on the negatve-class sub-database [5]. K-means fnds K locally optmal center objects by repeatedly applyng the EM algorthm on a set of data. Other good clusterng algorthms can also be used here n place of K-means. Step 6 of the algorthm fnds centrods of the K clusters found n the prevous step. These centrods are the bases of the case base constructed thus far, and are returned to the user. Fnally, Step 9 returns the case base as the output. The centrod-based case-mnng method extracts cases from the negatve-class cluster center objects and takes nto account only the negatve class dstrbuton. By consderng the dstrbuton of both the postve and negatve class clusters, we can sometmes do better. In general, t may be better to fnd the boundary between the two dstrbutons, and to select cases from the dense areas on that boundary. Ths s the ntuton behnd the SVM-based case mnng algorthm. The key ssue then s to dentfy the cases on the boundary between the postve and negatve cases, and select those cases as the fnal ones for the problem descrptons. Ths has two advantages. Frst, by usng the boundary objects as cases, from the past plans, we may determne that some objects are n fact not convertble to the postve class; these are the objects that are not smlar to any of the problem descrptons n the case base. The case-based reasonng algorthm cannot solve these cases. Second, 4

5 because the cases are the support vectors themselves, there s no need to specfy the nput parameter K as n the Centrods-CBMne algorthm; the parameter K s used to determne the number of clusters to be generated n K-means. Ths corresponds to parameter-less case mnng, whch s superor because n realty, the number of cases to be generated s typcally unknown ahead of tme. We observe that the cases along the boundary hyper-plane correspond to the support vectors found by an SVM classfer [7, 13]. These cases are the nstances that are closest to the maxmum margn hyper-plane n a hyperspace after an SVM system has dscovered the classfer [7, 13]. By explotng the above dea, we have a second problem-descrpton mnng algorthm called SVM-CBMne(). In the frst step, we perform SVM learnng on the database to locate the support vectors. Then we fnd the support vectors and nsert them nto the case base. Ths algorthm s llustrated n Table 4. Table 4. Algorthm SVM-CBMne (database DB, nt K) Steps Begn 1 casebase = empty set; 2 Vectors = SVM(DB); 3 for each negatve support vector C- n Vectors do 4 Insert(C-, casebase); 5 end for 6 Return casebase; End 4. Mnng the plans Havng obtaned the problem descrptons as seeds for the cases n the case base, we now consder the problem of fndng soluton plans to be assocated wth these descrptons wth an am to convert all negatve-class cases to postve. Our algorthm s a state-space search algorthm that search an AND-OR tree. AND-OR trees are used to deal wth the problem of uncertanty n plannng. The plans produced are probablstc n nature; they are amed at succeedng n convertng the customers to postve class wth low cost and hgh success probablty. Gven the customer table and the acton log database, our frst task s to formulate the problem as a plannng problem. In partcular, we wsh to fnd a method to map the customer records n the customer table nto states usng a statstcal classfer. Ths task n tself s not trval because t maps a large attrbute space nto a more concse space. When there are mssng values n the database, technques of data cleanng [14] can be appled to fll n these values. Next, the state-acton sequences n the acton log database wll be used for obtanng acton defntons n a state space, such that each acton s represented as a probablstc mappng from a state to a set of states. To make the representaton more realstc, we wll also consder the cost of executng each acton. 5

6 To summarze, from the two tables we can obtan the followng nformaton: f s ( ) = s j maps a customer record r to a state s j.ths functon s known as the customer-state mappng functon; p c ( s ) s a probablty functon that returns the probablty that state s s n a desrable class. We call ths classfer the state-classfcaton functon; p( sk s, a j ) returns the transton probablty that, after executng an acton a j n state s, one ends up n state s k. Our plannng algorthm dvdes the resultng plan n stages, where each stage conssts of one acton and a set of possble outcome states resultng from the acton. In each stage the states can be dfferent possble states as a result of the prevous acton, and the acton n the stage must be a same, sngle acton. In our formulaton, we defne a utlty functon of a plan to be u(s) for a state s, where the functon s defned as u( s) = max(( P( s' s, a)* u( s')) cost( a)) a s' S where, at a leaf node t, the functon u s defned as the utlty of the node whch s P(+ t), and cost(a) s the cost of the acton a. A major dffculty n solvng the plannng problem stems from the fact that there are potentally many states and many connectons between states. Ths potentally large space can be reduced sgnfcantly by observng that the states and ther connectons are not all equal; some states and acton sequences n ths state-space are more sgnfcant than others because they are more frequently traveled by traces n the actonlog table. Ths observaton allows us to use an approach n whch we explot plannng by abstracton. In partcular, sgnfcant state-acton sequences n the state space can be dscovered through a frequent strng-mnng algorthm [2]. We start by defnng a mnmumsupport threshold for fndng the frequent state-acton sequences. Support represents the number of occurrences of a state-acton sequence from the acton log database. More formally, let count(seq) be the number of tmes sequence seq appears n the database for all customers. Then the support for sequence seq s defned as the frequency of the sequence. Then, a strng-mnng algorthm based on movng wndows wll mne the acton log database to produce state-acton subsequences whose support s no less than a user-defned mnmum-support value. For connecton purpose, we only retan substrngs both begnnng and endng wth states, n the form of < s, a, s+ 1, a+ 1,..., sn >. Once the frequent sequences are found, we pece together the segments of paths correspondng to the sequences to buld an abstract AND-OR graph n whch we wll search for plans. If < s 0, a1, s2 > and < s 2, a3, s4 > are two segments found by the strng-mnng algorthm, then < s 0, a1, s2, a3, s4 > s a new path n the AND-OR graph. Snce each component of the AND-OR graph s guaranteed to be frequent, the AND-OR graph s a hghly concse and representatve state space. Based on the above heurstc estmaton methods, we can perform a best-frst search n the space of plans untl the termnaton condton s met. The termnaton condtons are determned by the probablty or the length constrants n the problem doman. For the ntal state s, f the expected value E(+ s) exceeds a predefned threshold Success_Threshold,.e. the probablty constrant, we consder the plan to be good enough and the search process termnates. Otherwse, one more acton s attached to ths plan and the new plans are nserted nto the prorty queue. E ( + s ) s the expected state-classfcaton probablty estmatng how effectve a plan s at transfer- 6

7 rng customers from state s. Its calculaton can be defned n the followng recursve way: E + s ) = p( s s, a )* E( + s ) ; f s s a non-termnal state; or ( k j k E + s ) = P( + s ) f s s a termnal state. ( We also defne a parameter Max_Step that defnes the maxmum length of a plan,.e. the length constrant. We wll dscard a canddate plan, whch s longer than the Max_Step but ts E + s ) value s less than the Success_Threshold. We now consder ( optmalty of the algorthm. 5 Effcency Experments We run tests on both artfcal data sets and realstc data sets to obtan the performance of the case mnng algorthm. In ths paper, we are prmarly nterested n the speed n whch to obtan the case bases, especally when the database reaches a large sze. Our other ongong work s amed at showng the qualty of the mned case bases. Thus, our nterest n these tests s how the CPU tme changes wth dfferent expermental parameters. In the sequel, we separately test the problem descrpton and the plannng algorthms. The experments are performed on an Intel PC wth one Ggahertz CPU. 5.1 Testng Problem Descrpton Mnng Our frst test s amed at establshng the effcency of the problem-descrpton mnng algorthm, whch determnes whether the case mnng framework can be scaled up or not. It uses an artfcal dataset generated on a two-dmensonal space (x, y), usng a Gaussan dstrbuton wth dfferent means and co-varance matrx for the postve (+) and the negatve ( ) classes (Fgure 1). Our purpose s to demonstrate the effect of data dstrbuton and model sze on the swtchng-plan qualty and effcency. When the means of the two dstrbutons are separated, we expect the class boundares are easy to dentfy by the SVM-based method. As can be seen, the tme t takes to buld and execute the model ncreases wth K, the number of cases n the case base (Fgure 2). As the two dstrbutons move close to each other such that there are no clear boundares, the SVM-based method selects nearly all the negatve examples as cases n the case base, resultng n a bloated case-base. Thus, ts tme expense s also very hgh (Fgure 2). In ths case, the CenrodCaseMne method s preferred. 7

8 Fgure 1. Dstrbuton of the class 1=postve (+) and class 2=negatve (*) data. CPU Seconds K=10 K=20 K=30 K=40 K=50 K=60 Sze of CaseBase K=70 K=80 K=90 K=100 SVM baselne Far Close Fgure 2. Comparson of CPU tme for dfferent dstrbutons between two dstrbutons that are ether Far apart or Closely mxed. For large-scale tests, we performed an experment usng the IBM QUEST synthetc data generator ( We generated the tranng dataset wth 9 attrbutes, 50% postve class and 50% negatve class. An excerpt of the database s shown n Table 5. In the dataset, the Salary attrbute s unformly dstrbuted from to , the commsson values are set so that f Salary >= 75000, Commsson = 0; otherwse t s unformly dstrbuted from to The Age attrbute s unformly dstrbuted from 20 to 80, the Educaton attrbute s unformly chosen from 0 to 4, and the Car attrbute ncludes the make of the car, and s unformly chosen from 1 to 20. 8

9 The ZpCode attrbute s unformly chosen from 9 avalable zp codes, and the House Value attrbute s unformly dstrbuted from 0.5*k* to 1.5*k*100000, where 0 <= k <= 9 and the House Value depends on the ZpCode. The YearsOwned attrbute s unformly dstrbuted from 1 to 30, and the Loan attrbute s unformly dstrbuted from 0 to Then we compared the runnng tme to tran a 50-case case base for the centrod-based method and CPU tme for the SVM based method. Our results are shown n Table 6. It s clear from the table that wth large data, the centrod-based method s able to scale up much better than the SVM based method. Table 5. An excerpt from the synthetc dataset. Salary Commsson Age Educaton Car Table 6. CPU-tme comparson of Centrod-based Method and SVM-based method. Log 10 (N), CPU Tme (Seconds) N=Database sze Centrod-based SVM , No result n 5 hrs 5 1,938.4 No result n 7 hrs 5.2 Testng Problem Descrpton Mnng In ths second test, we assume that we have obtaned a case base of typcal problem descrptons from the negatve examples n the tranng set. Our task now s to fnd a plan for each negatve descrpton. These plans wll serve as the cases that can be adapted for new problems n the future. In ths experment, we test the effcency of the algorthm for a sngle problem descrpton n a case. We agan used the IBM Synthetc Generator to generate a Customer dataset wth two classes and nne attrbutes. The postve class has 30,000 records representng successful customers and negatve has 70,000 representng unsuccessful ones. Those 70,000 negatve records are treated as startng ponts for Marketng-log data genera- 9

10 ton. We evaluated the qualty of the plans va smulaton. From ths nput, we set out to fnd a case base of marketng plans to convert customers to a successful class. Ths testng process corresponds to the testng phase for a traned model n machne learnng. In ths smulaton, f there s a plan sutable for convertng a customer record, a sequence of actons s carred out on that record. The plan wll then change the customer record probablstcally. At the end, the classfer s used to decde whether the changed record has turned nto a successful one. When Success_Threshold s low, many states are consdered postve, and thus plans can be easly and quckly found for most of the ntal states n the graph. We can observe ths from Fgure 3: plannng tme s low wth low Success_Threshold. As the Success_Threshold ncreases, so does the plannng tme. When Success_Threshold s too hgh, no plan can be found for some ntal states. The Tme s much hgher because the searchng process doesn t termnate untl all the plans expanded longer than Max_Step. The search effcency also depends on other parameters. MnSupport s another mportant factor. When MnSupport s low, more plans n the acton log qualfy to be n the abstract state space, and thus search takes a longer tme. We can also see from Fgure 4 that the CPU tme drop s greater when the MnSupport ncreases from one to 100. However, the plannng tme drop slows down between Mn- Support value of 100 and Ths suggest to us that we should use a support value of around 100, snce at ths value we can acheve the balance of a relatvely large number of plans to be ncluded n the search space and a low plannng tme. Tme(s) Fgure 3. CPU Tme vs. Success_Threshold. MnSupport = Tme(s) Fgure 4. CPU Tme vs. MnSupport. Success_Threshold=

11 6 Conclusons and Future Work In ths paper, we presented a case mnng algorthm for dscoverng a case base from a large database. The central ssue of the problem les n the dscovery of hgh-qualty case bases that balances the effcency for extractng the case base, the cost of the adaptaton plans and the probablty of success. Ths s what we call the case-mnng problem. For the data dstrbuton where the two classes are clearly separated, the SVM-CBMne algorthm, whch s an SVM-based method, should be used. When the data dstrbutons are not separated well by a boundary, the cluster-centrods based method s recommended. Furthermore, we desgned a plannng system for fndng probablstc plans where the probablty of success s taken nto account. The effcency of the case mnng algorthm s valdated aganst large databases. In the future, we wll contnue to explore qualty of case mnng results and the related problem of case adaptaton n ths framework. References [1] A. Aamodt, H. A. Sandtorv, O. M. Wnnem: Combnng Case Based Reasonng and Data Mnng - A way of revealng and reusng RAMS experence. In Lydersen, Hansen, Sandtorv (eds.), Safety and Relablty; Proceedngs of ESREL 98, Trondhem, June 16-19, Balkena, Rotterdam, [2] R. Agrawal and R. Srkant Fast algorthm for mnng assocaton rules. Proceedngs of the Twenteth Internatonal Conference on Very Large Databases. pp [3] C. L. Blake, C.J. Merz (1998). UCI Repostory of machne learnng databases Irvne, CA: Unversty of Calforna, Department of Informaton and Computer Scence. [4] A. Bonzano, P. Cunnngham and B. Smyth Usng Introspectve Learnng to Improve Retreval n CBR: A Case Study n Ar Traffc Contol, n Leake D., Plaza E. (eds.) `Case-Based Reasonng Research and Development, Second Internatonal Conference on Case-Based Reasonng ICCBR 1997, Sprnger Verlag, Berln [5] P. S. Bradley and U. M. Fayyad. Refnng ntal ponts for k-means clusterng. In Proceedngs of the Ffteenth Internatonal Conference on Machne Learnng (ICML '98), pages , San Francsco, CA, Morgan Kaufmann. [6] Computatonal Intellgence Journal, Specal Issue on Case-base Mantenance. Blackwell Publshers, Boston MA UK. Vol. 17, No. 2, May Edtors: D. Leake, B. Smyth, D. Wlson and Q. Yang. [7] G. C. Cowley. MATLAB Support Vector Machne Toolbox. v0.54b Unversty of East Angla, School of Informaton Systems, Norwch, Norfolk, U.K. NR4 7TJ,

12 [8] P. Domngos and M. Rchardson. Mnng the Network Value of Customers. Proceedngs of the Seventh ACM SIGKDD Internatonal Conference on Knowledge Dscovery and Data Mnng. August ACM. N.Y. N.Y. USA [9] K. Hammond. Explanng and Reparng Plans that Fal. Artfcal Intellgence, 45(1-2): , [10] D. Leake. Case-based Reasonng -- Experences, Lessons and Future Drectons. AAAI Press/ The MIT Press, [11] D. Leake, Knley, A., & Wlson, D. (1995). Learnng to mprove case adaptaton by ntrospectve reasonng and CBR. In Proceedngs of Frst Internatonal Conference on Case-Based Reasonng. Sesmbra, Portugal. [12] C. X. Lng and C. L. Data mnng for drect marketng: Problems and solutons. In Proceedngs 4th Internatonal Conference on Knowledge Dscovery n Databases (KDD-98), New York, [13] J.C. Platt, Fast tranng of support vector machnes usng sequental mnmal optmzaton, n Advances n Kernel Methods - Support Vector Learnng, (Eds) B. Scholkopf, C. Burges, and A. J. Smola, MIT Press, Cambrdge, Massachusetts, chapter 12, pp , [14] J. Qunlan C4.5: Programs for Machne Learnng. Morgan Kaufmann Publshers, Inc., San Mateo, CA. [15] B. Smyth and M. T. Keane. Rememberng to forget: A competence--preservng deleton polcy for case--based reasonng systems. In Proceedngs of the 14th Internatonal Jont Conference on Artfcal Intellgence, pp , 1995 [16] Q. Yang and J. Zhu. A Case-addton Polcy for Case-based Reasonng. Computatonal Intellgence Journal, Specal Issue on Case-based Mantenance. Blackwell Publshers, Boston MA UK. Vol. 17, No. 2, May Pages [17] D. C. Wlson and D. B. Leake Mantanng Case-based Reasoners: Dmensons and Drectons. Computatonal Intellgence Journal. Vol. 17, No. 2. May 2001 [18] M. Veloso. (1994). Plannng and learnng by analogcal reasonng. Number 886 n Lecture Notes n Artfcal Intellgence. Sprnger Verlag. [19] D. Wettschereck and D.V. Aha. Weghtng features. In Proceedngs of the Frst Internatonal Conference on Case-Based Reasonng, ICCBR-95, pages , Lsbon, Portugal, Sprnger-Verlag. [20] Zhong Zhang and Qang Yang. Feature Weght Mantenance n Case Bases Usng Introspectve Learnng. Journal of Intellgent Informaton Systems, Kluwer Academc Publshers, 16, Pages , The Netherlands. [21] D. Patterson, N. Rooney, M. Galushka, S. S. Anand: Towards Dynamc Mantenance of Retreval Knowledge n CBR. In Proceedngs of the Ffteenth Internatonal Florda Artfcal Intellgence Research Socety Conference, 2002, Florda, USA. AAAI Press 2002, pages

13 [22] B. Smyth, and E. McKenna, Buldng Compact Competent Case-Bases. In Proceedngs of the thrd Internatonal Conference on Case-based Reasonng, Sprnger-Verlag, Munch, Germany, 1999, pp

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

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

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 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

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

Concurrent Apriori Data Mining Algorithms

Concurrent Apriori Data Mining Algorithms Concurrent Apror Data Mnng Algorthms Vassl Halatchev Department of Electrcal Engneerng and Computer Scence York Unversty, Toronto October 8, 2015 Outlne Why t s mportant Introducton to Assocaton Rule Mnng

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

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

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

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

Outline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1

Outline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1 4/14/011 Outlne Dscrmnatve classfers for mage recognton Wednesday, Aprl 13 Krsten Grauman UT-Austn Last tme: wndow-based generc obect detecton basc ppelne face detecton wth boostng as case study Today:

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

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

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

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

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

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

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms Course Introducton Course Topcs Exams, abs, Proects A quc loo at a few algorthms 1 Advanced Data Structures and Algorthms Descrpton: We are gong to dscuss algorthm complexty analyss, algorthm desgn technques

More information

Problem Set 3 Solutions

Problem Set 3 Solutions Introducton to Algorthms October 4, 2002 Massachusetts Insttute of Technology 6046J/18410J Professors Erk Demane and Shaf Goldwasser Handout 14 Problem Set 3 Solutons (Exercses were not to be turned n,

More information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

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

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr)

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr) Helsnk Unversty Of Technology, Systems Analyss Laboratory Mat-2.08 Independent research projects n appled mathematcs (3 cr) "! #$&% Antt Laukkanen 506 R ajlaukka@cc.hut.f 2 Introducton...3 2 Multattrbute

More information

Fuzzy Modeling of the Complexity vs. Accuracy Trade-off in a Sequential Two-Stage Multi-Classifier System

Fuzzy Modeling of the Complexity vs. Accuracy Trade-off in a Sequential Two-Stage Multi-Classifier System Fuzzy Modelng of the Complexty vs. Accuracy Trade-off n a Sequental Two-Stage Mult-Classfer System MARK LAST 1 Department of Informaton Systems Engneerng Ben-Guron Unversty of the Negev Beer-Sheva 84105

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

Machine Learning. Topic 6: Clustering

Machine Learning. Topic 6: Clustering Machne Learnng Topc 6: lusterng lusterng Groupng data nto (hopefully useful) sets. Thngs on the left Thngs on the rght Applcatons of lusterng Hypothess Generaton lusters mght suggest natural groups. Hypothess

More information

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009.

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009. Farrukh Jabeen Algorthms 51 Assgnment #2 Due Date: June 15, 29. Assgnment # 2 Chapter 3 Dscrete Fourer Transforms Implement the FFT for the DFT. Descrbed n sectons 3.1 and 3.2. Delverables: 1. Concse descrpton

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

EECS 730 Introduction to Bioinformatics Sequence Alignment. Luke Huan Electrical Engineering and Computer Science

EECS 730 Introduction to Bioinformatics Sequence Alignment. Luke Huan Electrical Engineering and Computer Science EECS 730 Introducton to Bonformatcs Sequence Algnment Luke Huan Electrcal Engneerng and Computer Scence http://people.eecs.ku.edu/~huan/ HMM Π s a set of states Transton Probabltes a kl Pr( l 1 k Probablty

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

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

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

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

FEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur

FEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur FEATURE EXTRACTION Dr. K.Vjayarekha Assocate Dean School of Electrcal and Electroncs Engneerng SASTRA Unversty, Thanjavur613 41 Jont Intatve of IITs and IISc Funded by MHRD Page 1 of 8 Table of Contents

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

A Deflected Grid-based Algorithm for Clustering Analysis

A Deflected Grid-based Algorithm for Clustering Analysis A Deflected Grd-based Algorthm for Clusterng Analyss NANCY P. LIN, CHUNG-I CHANG, HAO-EN CHUEH, HUNG-JEN CHEN, WEI-HUA HAO Department of Computer Scence and Informaton Engneerng Tamkang Unversty 5 Yng-chuan

More information

BAYESIAN MULTI-SOURCE DOMAIN ADAPTATION

BAYESIAN MULTI-SOURCE DOMAIN ADAPTATION BAYESIAN MULTI-SOURCE DOMAIN ADAPTATION SHI-LIANG SUN, HONG-LEI SHI Department of Computer Scence and Technology, East Chna Normal Unversty 500 Dongchuan Road, Shangha 200241, P. R. Chna E-MAIL: slsun@cs.ecnu.edu.cn,

More information

X- Chart Using ANOM Approach

X- Chart Using ANOM Approach ISSN 1684-8403 Journal of Statstcs Volume 17, 010, pp. 3-3 Abstract X- Chart Usng ANOM Approach Gullapall Chakravarth 1 and Chaluvad Venkateswara Rao Control lmts for ndvdual measurements (X) chart are

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

Hierarchical clustering for gene expression data analysis

Hierarchical clustering for gene expression data analysis Herarchcal clusterng for gene expresson data analyss Gorgo Valentn e-mal: valentn@ds.unm.t Clusterng of Mcroarray Data. Clusterng of gene expresson profles (rows) => dscovery of co-regulated and functonally

More information

A User Selection Method in Advertising System

A User Selection Method in Advertising System Int. J. Communcatons, etwork and System Scences, 2010, 3, 54-58 do:10.4236/jcns.2010.31007 Publshed Onlne January 2010 (http://www.scrp.org/journal/jcns/). A User Selecton Method n Advertsng System Shy

More information

User Authentication Based On Behavioral Mouse Dynamics Biometrics

User Authentication Based On Behavioral Mouse Dynamics Biometrics User Authentcaton Based On Behavoral Mouse Dynamcs Bometrcs Chee-Hyung Yoon Danel Donghyun Km Department of Computer Scence Department of Computer Scence Stanford Unversty Stanford Unversty Stanford, CA

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

Kent State University CS 4/ Design and Analysis of Algorithms. Dept. of Math & Computer Science LECT-16. Dynamic Programming

Kent State University CS 4/ Design and Analysis of Algorithms. Dept. of Math & Computer Science LECT-16. Dynamic Programming CS 4/560 Desgn and Analyss of Algorthms Kent State Unversty Dept. of Math & Computer Scence LECT-6 Dynamc Programmng 2 Dynamc Programmng Dynamc Programmng, lke the dvde-and-conquer method, solves problems

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

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges

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

Collaboratively Regularized Nearest Points for Set Based Recognition

Collaboratively Regularized Nearest Points for Set Based Recognition Academc Center for Computng and Meda Studes, Kyoto Unversty Collaboratvely Regularzed Nearest Ponts for Set Based Recognton Yang Wu, Mchhko Mnoh, Masayuk Mukunok Kyoto Unversty 9/1/013 BMVC 013 @ Brstol,

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

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

Incremental Learning with Support Vector Machines and Fuzzy Set Theory

Incremental Learning with Support Vector Machines and Fuzzy Set Theory The 25th Workshop on Combnatoral Mathematcs and Computaton Theory Incremental Learnng wth Support Vector Machnes and Fuzzy Set Theory Yu-Mng Chuang 1 and Cha-Hwa Ln 2* 1 Department of Computer Scence and

More information

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster

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

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation 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 An Iteratve Soluton Approach to Process Plant Layout usng Mxed

More information

Intelligent Information Acquisition for Improved Clustering

Intelligent Information Acquisition for Improved Clustering Intellgent Informaton Acquston for Improved Clusterng Duy Vu Unversty of Texas at Austn duyvu@cs.utexas.edu Mkhal Blenko Mcrosoft Research mblenko@mcrosoft.com Prem Melvlle IBM T.J. Watson Research Center

More information

Private Information Retrieval (PIR)

Private Information Retrieval (PIR) 2 Levente Buttyán Problem formulaton Alce wants to obtan nformaton from a database, but she does not want the database to learn whch nformaton she wanted e.g., Alce s an nvestor queryng a stock-market

More information

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1)

For instance, ; the five basic number-sets are increasingly more n A B & B A A = B (1) Secton 1.2 Subsets and the Boolean operatons on sets If every element of the set A s an element of the set B, we say that A s a subset of B, or that A s contaned n B, or that B contans A, and we wrte A

More information

K-means and Hierarchical Clustering

K-means and Hierarchical Clustering Note to other teachers and users of these sldes. Andrew would be delghted f you found ths source materal useful n gvng your own lectures. Feel free to use these sldes verbatm, or to modfy them to ft your

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

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

MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION

MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION Paulo Quntlano 1 & Antono Santa-Rosa 1 Federal Polce Department, Brasla, Brazl. E-mals: quntlano.pqs@dpf.gov.br and

More information

Lecture 5: Multilayer Perceptrons

Lecture 5: Multilayer Perceptrons Lecture 5: Multlayer Perceptrons Roger Grosse 1 Introducton So far, we ve only talked about lnear models: lnear regresson and lnear bnary classfers. We noted that there are functons that can t be represented

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

CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION

CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION 48 CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION 3.1 INTRODUCTION The raw mcroarray data s bascally an mage wth dfferent colors ndcatng hybrdzaton (Xue

More information

Proper Choice of Data Used for the Estimation of Datum Transformation Parameters

Proper Choice of Data Used for the Estimation of Datum Transformation Parameters Proper Choce of Data Used for the Estmaton of Datum Transformaton Parameters Hakan S. KUTOGLU, Turkey Key words: Coordnate systems; transformaton; estmaton, relablty. SUMMARY Advances n technologes and

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

A Modified Median Filter for the Removal of Impulse Noise Based on the Support Vector Machines

A Modified Median Filter for the Removal of Impulse Noise Based on the Support Vector Machines A Modfed Medan Flter for the Removal of Impulse Nose Based on the Support Vector Machnes H. GOMEZ-MORENO, S. MALDONADO-BASCON, F. LOPEZ-FERRERAS, M. UTRILLA- MANSO AND P. GIL-JIMENEZ Departamento de Teoría

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

Optimal Workload-based Weighted Wavelet Synopses

Optimal Workload-based Weighted Wavelet Synopses Optmal Workload-based Weghted Wavelet Synopses Yoss Matas School of Computer Scence Tel Avv Unversty Tel Avv 69978, Israel matas@tau.ac.l Danel Urel School of Computer Scence Tel Avv Unversty Tel Avv 69978,

More information

Skew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach

Skew Angle Estimation and Correction of Hand Written, Textual and Large areas of Non-Textual Document Images: A Novel Approach Angle Estmaton and Correcton of Hand Wrtten, Textual and Large areas of Non-Textual Document Images: A Novel Approach D.R.Ramesh Babu Pyush M Kumat Mahesh D Dhannawat PES Insttute of Technology Research

More information

Unsupervised Learning and Clustering

Unsupervised Learning and Clustering Unsupervsed Learnng and Clusterng Why consder unlabeled samples?. Collectng and labelng large set of samples s costly Gettng recorded speech s free, labelng s tme consumng 2. Classfer could be desgned

More information

Parallel matrix-vector multiplication

Parallel matrix-vector multiplication Appendx A Parallel matrx-vector multplcaton The reduced transton matrx of the three-dmensonal cage model for gel electrophoress, descrbed n secton 3.2, becomes excessvely large for polymer lengths more

More information

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

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 New Token Allocation Algorithm for TCP Traffic in Diffserv Network

A New Token Allocation Algorithm for TCP Traffic in Diffserv Network A New Token Allocaton Algorthm for TCP Traffc n Dffserv Network A New Token Allocaton Algorthm for TCP Traffc n Dffserv Network S. Sudha and N. Ammasagounden Natonal Insttute of Technology, Truchrappall,

More information

Enhancement of Infrequent Purchased Product Recommendation Using Data Mining Techniques

Enhancement of Infrequent Purchased Product Recommendation Using Data Mining Techniques Enhancement of Infrequent Purchased Product Recommendaton Usng Data Mnng Technques Noraswalza Abdullah, Yue Xu, Shlomo Geva, and Mark Loo Dscplne of Computer Scence Faculty of Scence and Technology Queensland

More information

Efficient Text Classification by Weighted Proximal SVM *

Efficient Text Classification by Weighted Proximal SVM * Effcent ext Classfcaton by Weghted Proxmal SVM * Dong Zhuang 1, Benyu Zhang, Qang Yang 3, Jun Yan 4, Zheng Chen, Yng Chen 1 1 Computer Scence and Engneerng, Bejng Insttute of echnology, Bejng 100081, Chna

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

Yan et al. / J Zhejiang Univ-Sci C (Comput & Electron) in press 1. Improving Naive Bayes classifier by dividing its decision regions *

Yan et al. / J Zhejiang Univ-Sci C (Comput & Electron) in press 1. Improving Naive Bayes classifier by dividing its decision regions * Yan et al. / J Zhejang Unv-Sc C (Comput & Electron) n press 1 Journal of Zhejang Unversty-SCIENCE C (Computers & Electroncs) ISSN 1869-1951 (Prnt); ISSN 1869-196X (Onlne) www.zju.edu.cn/jzus; www.sprngerlnk.com

More information

Design and Analysis of Algorithms

Design and Analysis of Algorithms Desgn and Analyss of Algorthms Heaps and Heapsort Reference: CLRS Chapter 6 Topcs: Heaps Heapsort Prorty queue Huo Hongwe Recap and overvew The story so far... Inserton sort runnng tme of Θ(n 2 ); sorts

More information

Japanese Dependency Analysis Based on Improved SVM and KNN

Japanese Dependency Analysis Based on Improved SVM and KNN Proceedngs of the 7th WSEAS Internatonal Conference on Smulaton, Modellng and Optmzaton, Bejng, Chna, September 15-17, 2007 140 Japanese Dependency Analyss Based on Improved SVM and KNN ZHOU HUIWEI and

More information

TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS. Muradaliyev A.Z.

TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS. Muradaliyev A.Z. TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS Muradalyev AZ Azerbajan Scentfc-Research and Desgn-Prospectng Insttute of Energetc AZ1012, Ave HZardab-94 E-mal:aydn_murad@yahoocom Importance of

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

Fast Feature Value Searching for Face Detection

Fast Feature Value Searching for Face Detection Vol., No. 2 Computer and Informaton Scence Fast Feature Value Searchng for Face Detecton Yunyang Yan Department of Computer Engneerng Huayn Insttute of Technology Hua an 22300, Chna E-mal: areyyyke@63.com

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

Elliptical Rule Extraction from a Trained Radial Basis Function Neural Network

Elliptical Rule Extraction from a Trained Radial Basis Function Neural Network Ellptcal Rule Extracton from a Traned Radal Bass Functon Neural Network Andrey Bondarenko, Arkady Borsov DITF, Rga Techncal Unversty, Meza ¼, Rga, LV-1658, Latva Andrejs.bondarenko@gmal.com, arkadjs.borsovs@cs.rtu.lv

More information

Comparing High-Order Boolean Features

Comparing High-Order Boolean Features Brgham Young Unversty BYU cholarsarchve All Faculty Publcatons 2005-07-0 Comparng Hgh-Order Boolean Features Adam Drake adam_drake@yahoo.com Dan A. Ventura ventura@cs.byu.edu Follow ths and addtonal works

More information

CAN COMPUTERS LEARN FASTER? Seyda Ertekin Computer Science & Engineering The Pennsylvania State University

CAN COMPUTERS LEARN FASTER? Seyda Ertekin Computer Science & Engineering The Pennsylvania State University CAN COMPUTERS LEARN FASTER? Seyda Ertekn Computer Scence & Engneerng The Pennsylvana State Unversty sertekn@cse.psu.edu ABSTRACT Ever snce computers were nvented, manknd wondered whether they mght be made

More information

Sorting Review. Sorting. Comparison Sorting. CSE 680 Prof. Roger Crawfis. Assumptions

Sorting Review. Sorting. Comparison Sorting. CSE 680 Prof. Roger Crawfis. Assumptions Sortng Revew Introducton to Algorthms Qucksort CSE 680 Prof. Roger Crawfs Inserton Sort T(n) = Θ(n 2 ) In-place Merge Sort T(n) = Θ(n lg(n)) Not n-place Selecton Sort (from homework) T(n) = Θ(n 2 ) In-place

More information

A Saturation Binary Neural Network for Crossbar Switching Problem

A Saturation Binary Neural Network for Crossbar Switching Problem A Saturaton Bnary Neural Network for Crossbar Swtchng Problem Cu Zhang 1, L-Qng Zhao 2, and Rong-Long Wang 2 1 Department of Autocontrol, Laonng Insttute of Scence and Technology, Benx, Chna bxlkyzhangcu@163.com

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

Feature Selection as an Improving Step for Decision Tree Construction

Feature Selection as an Improving Step for Decision Tree Construction 2009 Internatonal Conference on Machne Learnng and Computng IPCSIT vol.3 (2011) (2011) IACSIT Press, Sngapore Feature Selecton as an Improvng Step for Decson Tree Constructon Mahd Esmael 1, Fazekas Gabor

More information

A Notable Swarm Approach to Evolve Neural Network for Classification in Data Mining

A Notable Swarm Approach to Evolve Neural Network for Classification in Data Mining A Notable Swarm Approach to Evolve Neural Network for Classfcaton n Data Mnng Satchdananda Dehur 1, Bjan Bhar Mshra 2 and Sung-Bae Cho 1 1 Soft Computng Laboratory, Department of Computer Scence, Yonse

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

Support Vector Machines. CS534 - Machine Learning

Support Vector Machines. CS534 - Machine Learning Support Vector Machnes CS534 - Machne Learnng Perceptron Revsted: Lnear Separators Bnar classfcaton can be veed as the task of separatng classes n feature space: b > 0 b 0 b < 0 f() sgn( b) Lnear Separators

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

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements Module 3: Element Propertes Lecture : Lagrange and Serendpty Elements 5 In last lecture note, the nterpolaton functons are derved on the bass of assumed polynomal from Pascal s trangle for the fled varable.

More information

Parameter estimation for incomplete bivariate longitudinal data in clinical trials

Parameter estimation for incomplete bivariate longitudinal data in clinical trials Parameter estmaton for ncomplete bvarate longtudnal data n clncal trals Naum M. Khutoryansky Novo Nordsk Pharmaceutcals, Inc., Prnceton, NJ ABSTRACT Bvarate models are useful when analyzng longtudnal data

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

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