D. Barbuzzi* and G. Pirlo

Size: px
Start display at page:

Download "D. Barbuzzi* and G. Pirlo"

Transcription

1 Int. J. Sgnal and Imagng Systems Engneerng, Vol. 7, No. 4, bout retranng rule n mult-expert ntellgent system for sem-supervsed learnng usng SVM classfers D. Barbuzz* and G. Prlo Department of Computer Scence, Unversty of Bar ldo Moro, Bar, Italy Emal: donato.barbuzz@unba.t Emal: guseppe.prlo@unba.t *Correspondng author D. Impedovo Department of Electrcal and Electronc Engneerng, Polytechnc of Bar, Bar, Italy Emal: mpedovo@deemal.polba.t bstract: Ths paper proposes three methods n order to retran classfers n a mult-expert scenaro, when new (unknown) data are avalable. In fact, when a mult-expert system s adopted, the collectve behavour of classfers can be used for both recognton ams and selecton of the most proftable samples for system retranng. More specfcally a msclassfed sample for a partcular expert can be used to update the expert tself f the collectve behavour of the multexpert system allows the classfcaton of the sample wth hgh confdence. In addton, ths paper provdes a comparson between the new approach and those avalable n the lterature for semsupervsed learnng usng the SVM classfer by takng nto account four dfferent combnaton technques at abstract and measurement levels. The expermental results, whch have been obtaned usng the handwrtten dgts of the CEDR database, demonstrate the effectveness of the proposed approach. Keywords: feedback-based strateges; sem-supervsed learnng; ntellgent mult-expert system. Reference to ths paper should be made as follows: Barbuzz, D., Prlo, G. and Impedovo, D. (2014) bout retranng rule n mult-expert ntellgent system for sem-supervsed learnng usng SVM classfers, Int. J. Sgnal and Imagng Systems Engneerng, Vol. 7, No. 4, pp Bographcal notes: D. Barbuzz receved the Computer Scence degree cum laude n 2011 from the Unversty of Bar ldo Moro. He worked from September to December 2011 as Collaborator n the Interfaculty Center Rete Pugla. Snce 2012, he has been a PhD student at the Department of Computer Scence at the Unversty of Bar. Hs current research nterest s n the feld of mult-expert systems for pattern recognton. G. Prlo receved the Computer Scence degree cum laude n 1986 at the Unversty of Bar, Italy. Snce 1991 he has been ssstant Professor at the same Unversty, where he s currently ssocate Professor. Hs nterests cover the areas of pattern recognton, bometry, computer arthmetc, communcaton and multmeda technologes. He has developed several scentfc projects and publshed over 200 papers. He s assocate edtor of IEEE T-HMS and revewer for T-PMI, T-SMC, T-IP, PR, IJDR, IPL, etc. He was the general co-char of the ICFHR 2012 and EHSP He s IEEE and IPR member. D. Impedovo receved the MEng degree cum laude n Computer Engneerng n 2005 and the PhD degree n Computer Engneerng n 2009 both from the Polytechnc of Bar (Italy). In 2011 he receved the MSc on Remote Scence Technologes from Bar Unversty. Hs research nterests are n the feld of pattern recognton and bometrcs. He s co-author of more than 20 artcles n both nternatonal journals and conference proceedngs. He receved The Dstncton for the best young student presentaton n May 2009 at the Internatonal Conference on Computer Recognton Systems (CORES endorsed by IPR). He s IPR and IEEE member. Copyrght 2014 Inderscence Enterprses Ltd.

2 246 D. Barbuzz, G. Prlo and D. Impedovo 1 Introducton pattern recognton system s based on two man processes: enrolment (or tranng) and matchng (or recognton). In the enrolment process, samples of specfc classes are acqured and processed n order to extract relevant features. The features are labelled wth the ground truth and used to generate the model representng the class. However, t may be expensve n terms of space and tme to obtan a large amount of labelled data. On the other hand, t s relatvely easy to collect many unlabelled data that may be used to enlarge the labelled tranng set. Matchng process performs the recognton of the (unknown) nput pattern by comparng t to the enrolled data. Dependng on the specfc scenaro, t has been observed that the recognton accuracy can also be mproved by multexpert fuson (Kttler et al., 1998; Plamondon and Srhar, 2000; Suen and Tan, 2005; Lu et al., 2003; Impedovo et al., 2011; Impedovo et al., 2008; Impedovo et al., 2012; Barbuzz et al., 2012). For the purpose, many approaches have been proposed so far for classfers combnaton, whch dffer n terms of the type of output they combne, system topology and degree of a pror knowledge they use (Kttler et al., 1998; Plamondon and Srhar, 2000; Suen et al., 1992). Furthermore, when a Mult-Expert (ME) system s adopted, as new (unknown) data become avalable, the collectve behavour of classfers can be used not only to recognse patterns wth hgh accuracy but also to select the most proftable samples for updatng the knowledge base of the ndvdual classfers. Ths s partcularly true n supervsed learnng. s already dscussed n the lterature, the achevement of better performance depends on the teraton of the feedback process (Barbuzz et al., 2013), on the combnaton strategy of the ME, but also on data dstrbuton and the smlarty between samples n the feedback set and samples of the testng set (Prlo et al., 2009; Impedovo and Prlo, 2011; Impedovo et al., 2012; Barbuzz et al., 2012; Barbuzz et al., 2013). Ths paper presents a new feedback-based strategy for parallel topology. The basc motvaton of ths work s that the collectve behavour of a set of classfers generally conveys more nformaton than that of each classfer of the set, and ths nformaton can be exploted for system retranng (Suen and Tan, 2005; Prlo and Impedovo, 2011; Prlo et al., 2009; Impedovo and Prlo, 2011). More precsely, as Fgure 1 shows, the approach selects samples that the mult-expert system has classfed wth hgh confdence for updatng the ndvdual experts who have msclassfed the samples themselves. The expermental results, carred out on the handwrtten dgts of the CEDR database, demonstrate the effectveness of the proposed feedback-based strategy also wth respect to the algorthms self-tranng and co-tranng already presented n the lterature (Frnken and Bunke, 2009; Frnken et al., 2011; Blum and Mtchell, 1998; Rattan et al., 2009; Guo et al., 2011). In addton, the expermental results allow us to defne the extent to whch the performance of the new strategy depends on the specfc schema of combnaton decson as well as on data dstrbuton. The paper s organsed as follows. Secton 2 presents the technques of retranng n sem-supervsed learnng. Secton 3 descrbes n detal the new feedback-based strategy. The operatng condtons are presented n Secton 4. The expermental results are reported n Secton 5. Secton 6 presents the concluson of the work and a bref dscusson on future drectons of research. 2 Related work Let us consder a mult-expert system that s already traned and n ts workng phase. Let us suppose that new unlabelled data become avalable over tme; the queston s: how to use these new unknown data? The labellng method s a laborous and costly process. So, new unlabelled data can be used n order to enlarge the tranng set for mprovng the performance of the whole system by sem-supervsed learnng technques. In partcular, two very well known approaches n the lterature for ths purpose are (Frnken and Bunke, 2009; Frnken et al., 2011; Blum and Mtchell, 1998; Rattan et al., 2009; Guo et al., 2011): selftranng and co-tranng. Self-tranng (or self-update; Frnken and Bunke, 2009; Rattan et al., 2009; Guo et al., 2011), as shown n Fgure 2, s based on the concept that a classfer s retraned on ts own most confdent output produced from unlabelled data. The classfer s frst traned on the set of labelled data and, subsequently, several of the new unlabelled nstances that the current classfer has hgh classfcaton confdence are labelled and moved to enlarge ts tranng set. The whole process terates untl some stop rule s satsfed. Co-tranng (or co-update; Blum and Mtchell, 1998; Rattan et al., 2009; Guo et al., 2011), as shown n Fgure 2. Fgure 2 Flow chart for self-update approach (see onlne verson for colours) Fgure 1 x Feedback n the mult-expert parallel system 1 (x t ) 2 (x t ) E(x t ) E (x t ) N (x t ) Fgure 3, s the stuaton under whch two classfers mprove each other. More specfcally, the features set (vew) s parttoned nto two condtonally ndependent subsets (sub-

3 bout retranng rule n mult-expert ntellgent system 247 vews), so that for a gven class the features used for one classfer must not be correlated wth the features used for the other classfer. In ths manner, the frst expert s updated wth elements confdently recognsed by the other one and vce versa. The process terates untl stopped. Fgure 3 Flow chart for co-update approach (see onlne verson for colours) determnes whether or not the pattern s fed nto the system for updatng the knowledge base of ndvdual classfers. Fgure 4 x Self-update n the mult-expert parallel system 1 (x t ) 2 (x t ) (x t ) E E(x t ) N (x t ) The stoppng crteron n self-tranng and co-tranng s that ether there are no unlabelled nstances left or the maxmum number of teratons has been reached (Guo et al., 2011). Co-tranng and self-tranng have a very strong role, n the current state of the art, n bometrcs template updatng process (Uludag et al., 2004; Rattan et al., 2009); moreover, they have been appled recently even n the feld of handwrtng recognton (Frnken and Bunke, 2009; Frnken et al., 2011). The man result observed for self-tranng on the task of handwrtten word recognton (Frnken and Bunke, 2009) s that the challenge of successful self-tranng les n fndng the optmal trade-off between data qualty and data quantty for retranng. In partcular, f the retranng s done wth only those elements whose correctness can nearly be guaranteed, the retranng set does not change sgnfcantly and the classfer may reman nearly the same, or n other cases t could dscard genune samples whose dstrbuton s far from the one already embedded n the knowledge base, thus resultng n a performance degradaton. Enlargng the retranng set, on the other hand, s only possble at the cost of ncreasng nose,.e. addng mslabelled words to the tranng set. In ths scenaro, the cotranng approach (Frnken et al., 2011) appears to be much more nterestng; n fact, t does not suffer lmtatons of the self-update process, and performance mprovement are more evdent than those observed n the self-tranng case, even f the confdence threshold stll plays a crucal role. Co-tranng can be easly extended from two to n classfers, but the basc observaton s that, once more, even f an ensemble of classfers s avalable, there s no analyss and use of ther common behavour of classfcaton gven the nput to be recognsed. Self-update and co-update n mult-expert parallel system are nvestgated here (see Fgures 4 and 5, respectvely) and compared aganst our approach, where the fnal result E(xt) provded by the mult-expert system, accordng to (xt) provded by the sngle classfer, Fgure 5 x Co-update n the mult-expert parallel system 1 (x t ) 2 (x t ) (x t ) 3 Feedback-based strategy N (x t ) E E(x t ) The new feedback-based strategy for parallel topology s descrbed n the followng. 3.1 Instance selecton from ensemble behavour Let: Cj, for j = 1,2,, M: the set of pattern classes. P xk k 1,2,..., K : the set of pattern to be fed nto the Mult-Expert (ME) system. In ths work, P s consdered to be parttoned nto S subsets: PP,,, P,, P 1 2 s S where Ps xk P k Ns ( s 1) 1, Ns s and Ns K/ S, N s nteger, whch are provded one after the other to the mult-expert system. In partcular, P 1 s used only for learnng and ts data are labelled, whereas P2, P3,, Ps,, PS are used for both classfcaton and learnng (when necessary) and ther data are unlabelled. y s Ω: the label of the pattern x s, C1, C2,..., CM. : the th classfer, = 1,2,,N. F ( k) F k, F k,, F k, F k : the,1,2, r, R numeral feature vector used by for representng the pattern x k P (for the sake of smplcty, here t s assumed that each classfer uses R numeral features).

4 248 D. Barbuzz, G. Prlo and D. Impedovo KB (k): the knowledge base of after the processng of the subset P k. In partcular, 1 2 M,,, KB k KB k KB k KB k. E: the mult-expert system whch combnes the ndvdual classfer decsons n order to obtan the fnal classfcaton result. Intally, n the frst stage (s = 1), the classfer s traned * usng the patterns xk P P. Therefore, the knowledge 1 base KB (s) of s ntally defned as 1 2 j M,,,,, KB s KB s KB s KB s KB s (1) where for j = 1,2,, M, KB () s ( F (), s F (), s, F (), s, F ()) s (2) j j j j j,1,2, r, R j wth FR, () s beng the set of the rth feature of the th * classfer for the patterns of the class C j that belongs to P. Successvely, the subsets P 2, P 3,, Ps,, PS 1 of unknown samples are provded one after the other to the mult-classfer system both for classfcaton and for learnng. PS s just consdered to be the testng set n order to avod based or too optmstc results. When consderng new data (samples of P 2, P 3,, Ps,, PS 1 ), n order to nspect and take advantage of the common behavour of the ensemble of classfers, the followng smple strategy s proposed and evaluated n ths work: Ex Ext xt t and x t Ps ' : update _ KB scoreme threshold In other words, s updated wth the pattern that t msclassfes consderng the label, obtaned wth a hgh confdence measure, by ME system. It s worth notng that ths strategy (see equaton 3) takes nto account the performance of the ndvdual classfer as well as the performance of the ME system. It s able to select not only samples to be used for the updatng process but also classfers nto whch those samples must be fed n order to mprove the ME performance. 3.2 lgorthm In the feedback-based procedure, each expert s traned on an ntal set of enrolled data, named P 1. batch P f of (unknown) samples s collected over a certan tme. mong all the samples n P f, only those classfed wth hgh confdence by the mult-expert system are used to enlarge the knowledge base of those experts that produced a msclassfcaton. To compute the confdence threshold, P f s used as the valdaton set. The retranng rule computes both the confdence threshold Thr* for each expert and the confdence threshold ThrME for the mult-expert system. Successvely, all samples recognsed wth a confdence equal to or greater than ThrME and msclassfed by a specfc (3) expert are selected. Ths s done to reduce the probablty of mpostor ntroducton (false acceptance) nto the updated tranng set. Ths procedure s presented n lgorthm 1. lgorthm 1: Feedback-based Process 1 Gven: P1 p,, 1 p : M the Intal Tranng Set S -1 Pf Pu pm 1,, pn u2 : the new avalable (unknown) data set KB (k) : the knowledge base of the expert, =1,2,, N 2 For each expert: (a) (u, P 1 ) s the classfcaton of the sample u P f on P 1 (b) thr* s the estmated threshold for End for 3 pply the combnaton rule: E(u,P 1 ) s the classfcaton of the sample u P on P 1 combnng all experts. 4 thrme s the estmated threshold for E 5 Determne for each sample u Pf a classfcaton score of the Mult-Expert system sme Eu, P 1 6 For h = M + 1,, N 7 For each expert (a) If uh, P1 Euh, P1 sme Eu, P1 thrme and (b) Then KB ( k) KB ( k) u End for End for. h However, t s evdent that many new samples wll not be fed nto a specfc classfer and, n general, we could expect to observe performance degradaton f compared to other strateges. Of course, ths phenomenon depends on thresholds, classfer performances and the rato between the new and the old data. However, we have to consder and remark that we are dealng wth already traned and workng classfers, so that ther ntal performances are expected to be on the state of the art. Ths leads to two consderatons: 1 Gven a specfc classfer, the dfference between the confdence values n the case of msclassfcaton and n the case of correct classfcaton could be mputed to the fact that the specfc classfer s unable to represent a specfc class or sample, and no mprovements would be obtaned by ntroducng the new sample n the knowledge base. Ths s partcularly true under the assumpton that strong (not weak) classfers are used. 2 If each classfer n the ensemble were able to recognse exactly the same set of patterns, ther combnaton would be not useful (Impedovo et al., 2010). f

5 bout retranng rule n mult-expert ntellgent system Operatng condtons In ths paper, the handwrtten dgts P { x j 1,2,, 20351} (classes from 0 to 9 ) of the CEDR database (Hull, 1994) have been used. Fgure 6 presents some samples of handwrtten dgts. The DB has been ntally parttoned nto sx subsets: P x, x, x,,, x, P2 x12751,,, x14119, P3 x14120,,, x15488, P4 x15489,,, x16857, P5 x16858,,, x18223, P x,,, x In partcular, P1P2 P3 P4 P5 represents the set usually ad`opted for tranng when consderng the CEDR DB (Lu et al., 2003). P 6 s the testng data set. Each dgt s zoned nto 16 unform (regular) regons (Prlo and Impedovo, 2011); successvely for each regon the followng set of features were consdered (Lu et al., 2003): 1 Geometrc features: hole, up cavty, down cavty, left cavty, rght cavty, up end pont, down end pont, left end pont, rght end pont, crossng ponts, up extrema ponts, down extrema ponts, left extrema ponts, rght extrema ponts; 2 Contour profles: max/mn peaks, max/mn profles, max/mn wdth, max/mn heght; 3 Intersecton wth lnes: fve horzontal lnes, fve vertcal lnes, fve slant 45 lnes and fve slant +45 lnes. Fgure Thresholds Samples of handwrtten dgts of the CEDR database To estmate the qualty of the classfcaton output, addtonal nformaton about the classfcaton s necessary to ndcate the degree of confdence of the expert. To compute the confdence measure, the new set of avalable data s used at the same tme to classfy and retran, when necessary. More precsely, the retranng rules compared n ths work are based on ths confdence measure. The retranng rules, for self-tranng and co-tranng, defne a j specfc confdence threshold and all the dgts classfed wth a degree of confdence equal to or greater than ths threshold are added to the expert s knowledge base. The thrd rule, defned at the mult-expert level, selects all the nstances msclassfed, for a partcular expert, and updates t only f that sample produces a hgh classfcaton confdence by the ensemble of experts. The frst two thresholds are set to 0.75 normalsed z- score. The thrd confdence measure s the average between the thresholds by each expert set to 0.75 normalsed z-score. 3.2 Classfer and combnaton technques The expermental system uses three dfferent classfers, named 1, 2 and 3. Each classfer uses a dfferent set of features: 1 uses the feature set no. 1, 2 uses the feature set no. 2 and 1 uses the feature set no. 3. ll classfers perform classfcaton by a Support Vector Machne (SVM). Of course, snce SVM s a bnary (two-class) classfer, mult-class classfcaton s here performed by combnng multple bnary SVMs. In fact, an M-class problem can be decomposed nto M bnary problems wth each separatng one class from all the others (Chang and Ln, 2011). The kernel functon adopted n ths work s the rbf gamma. The performances are certanly nfluenced by the number of features (gamma) and by the tolerance of classfcaton errors n learnng (C) (Lu et al., 2003; Chang and Ln, 2011). Despte the power of the classfer, n mult-expert system the combnaton technque plays a crucal role n the selecton of new patterns to be fed nto the classfer n the proposed approach. In ths work, the followng decson combnaton strateges have been consdered and compared: Majorty Vote (MV), Weghted Majorty Vote (WMV), Sum Rule (SR) and Product Rule (PR). MV consders labels provded by the ndvdual classfers. It s generally adopted f no knowledge s avalable about performance of classfers. WMV can be adopted by consderng weghts related to the performance of ndvdual classfers on a specfc data set (Polkar, 2007). In the case depcted n ths work, t seems to be more realstc. In fact, the behavour of classfers can be evaluated, for nstance, on the new avalable data set. In partcular, let ε be the error rate of the th classfer evaluated on the last avalable tranng set, the weght assgned to s w log 1, where. For the sake of smplcty, let us consder an ME 1 adoptng three base classfers combned by means of a smple majorty vote approach (see Fgure 7). In the case depcted, the knowledge base of 1 wth x wll be updated, whle the second and the thrd experts wll not be updated. In ths case, the 1 performance would be ncreased on the tranng set and on pattern smlar to x, mprovng the overall ME performance. Fgure 7 Examples of updatng requests 1 x y 2 x y E x y 3 x y

6 250 D. Barbuzz, G. Prlo and D. Impedovo Sum Rule (SR) and Product Rule (PR) take nto account the confdence of each ndvdual classfer gven the nput pattern and the dfferent classes (Kttler et al., 1998). Of course, before combnaton, confdence values provded by dfferent classfers were normalsed by means of z-score. 4 Expermental results In the expermental tests, results are reported n terms of Error Rate (ER). Moreover, for the dfferent learnng strateges the numbers of Selected Samples (SS) from set of new data are reported. The label X-feed refers to the use of the X modalty for the feedback process. Precsely MV, WMV, SR and PR are feedback at ME level adoptng, respectvely, the majorty vote, the weghted majorty vote, the sum rule and the product rule. Table 1 shows the results related to the frst test. In partcular, P 1 s used for tranng and P 6 for testng. P 2, P 3, P 4 and P 5 were ndependently used, one from the other; for feedback learnng, performances were evaluated for each set and the average was fnally reported. The frst column (No-feed) reports results related to the use of P 1 for tranng and of P 6 for testng, wthout applyng any feedback (selected samples are 0). The total amount of new samples s n ths case (about) the 10% of the number of samples of the ntal tranng set (P 1 ). In ths case, expermental results (see Table 1) demonstrate the effectveness of the mult-expert feedback strategy. In MVfeed, WMV-feed and PR-feed, an mprovement of, respectvely, 0.11%, 0.07%, 0.05% can be found compared to self-update technque, whle for MV-feed, WMV-feed, SR-feed and PRfeed, there s an mprovement of, respectvely, 0.12%, 0.07%, 0.04% and 0.01% respect to co-update technque. Despte the few selected samples by the ME, the performance of the whole system outperforms those related self-tranng and co-tranng strateges. Table 2 shows the results related to the second experment. Ths test ams at evaluatng the proposed approach under the condton n whch the number of new pattern s sgnfcant f compared to the ntal tranng one. In partcular, P 1 s used for tranng and P 6 for testng. P 2 P 3 P 4 P 5 s used for feedback learnng. In ths case, the total amount of new samples s 42.86% of the number of samples of the ntal tranng set (P 1 ). Once more, the frst column (No-feed) reports results related to the use of P 1 for tranng and of P 6 for testng, wthout applyng any feedback (selected samples are 0). The behavour observed consderng the use of SVM classfers s smlar to the prevous experment; n fact, MV-feed, WMV-feed and SR-feed, respectvely, saw an mprovement of 0.24%, 0.09% and 0.10% wth respect to the self-update technque, whle for MV-feed, WMV-feed, SR-feed and PR-feed, respectvely, an mprovement of 0.24%, 0.14%, 0.05% and 0.19% can be seen consderng the co-update approach. The results demonstrate that our approach outperforms other approaches n the lterature n terms of both error rate and selected samples requested for the updatng process, when new (unknown) data became avalable. Table 1 SVM feedback P 2, P 3, P 4, P 5 No-feed Self-update Co-update MV-feed WMV-feed SR-feed PR-feed ER ER SS ER SS ER SS ER SS ER SS ER SS MV WMV SR PR Table 2 SVM feedback P 2 P 3 P 4 P 5 No-feed Self-update Co-update MV-feed WMV-feed SR-feed PR-feed ER ER SS ER SS ER SS ER SS ER SS ER SS , , , MV WMV SR X 1.36 PR

7 bout retranng rule n mult-expert ntellgent system Conclusons and future works Unlabelled data are most often abundant, but obtanng labels s expensve or tme-consumng. Instead of smply labellng all the new (unknown) data or randomly selectng data to be labelled, ths paper shows the possblty of mprovng (selectng specfc nstances) the effectveness of a mult-classfer system by a sutable use of the nformaton extracted from the collectve behavour of the classfers. More precsely, when a new unlabelled data set becomes avalable, the fnal decson (obtaned by combnng the decsons of the ndvdual classfers) has been used to upgrade the knowledge base of the ndvdual classfers, when necessary, accordng to a feedback-based topology. The expermental results (n terms of error rate and accuracy n samples selecton) reported n ths paper demonstrate that the collectve behavour of a set of classfers provdes useful nformaton to mprove system performance and the effectveness of the new feedback-based strategy when compared to other sem-supervsed learnng algorthms n the lterature, as self-tranng and co-tranng. Future works can nspect the possblty of teratve retranng gven a set of new unlabelled samples as well as the possblty to evaluate the method on the varablty of confdence threshold. The applcablty of the strategy n the context of unsupervsed learnng wll be also consdered. References Barbuzz, D., Impedovo, D. and Prlo, G. (2012) Benchmarkng of update learnng strateges on dgt classfer systems, Proceedngs of the 13th Internatonal Conference on Fronters n Handwrtng Recognton, September, Captolo, Bar, Italy, pp Barbuzz, D., Impedovo, D., Mangn, F.M. and Prlo, G. (2013) Learnng teratve strateges n mult-expert systems usng SVMs for dgt recognton, n Petrosno,. (Ed.): Proceedngs of the 17th Internatonal Conference on Image nalyss and Processng, September, Naples, Italy, LNCS, Vol. 8156, pp Blum,. and Mtchell, T. (1998) Combnng labeled and unlabeled data wth co-tranng, Proceedngs of the 11th nnual Conference on Computatonal Learnng Theory, July, Madson, WI, pp Chang, C-C. and Ln, C-J. (2011) LIBSVM: a lbrary for support vector machnes, CM Transactons on Intellgent Systems and Technology, Vol. 2, pp.27:1 27:27. valable onlne at: Frnken, V. and Bunke, H. (2009) Evaluatng retranng rules for sem-supervsed learnng n neural network based cursve word recognton, Proceedngs of the IEEE 10th Internatonal Conference on Document nalyss and Recognton (ICDR2009), July, Catalona, Span, pp Frnken, V., Fscher,., Bunke, H. and Fornes,. (2011) Cotranng for handwrtten word recognton, Proceedngs of the IEEE Internatonal Conference on Document nalyss and Recognton (ICDR2011), September, Bejng, Chna, pp Guo, Y., Zhang, H. and Lu, X. (2011) Instance selecton n semsupervsed learnng, Proceedngs of the 24th Canadan Conference on rtfcal Intellgence, May, St. John s, Canada, pp Hull, J. (1994) database for handwrtten text recognton research, IEEE Transactons on Pattern nalyss and Machne Intellgence, Vol. 16, No. 5, pp Impedovo, D. and Prlo, G. (2011) Updatng knowledge n feedbackbased mult-classfer systems, Proceedngs of the IEEE of Internatonal Conference on Document nalyss and Recognton (ICDR2011), September, Bejng, Chna, pp Impedovo, D., Prlo, G. and Barbuzz, D. (2012) Supervsed learnng strateges n mult-classfer systems, Proceedngs of 11th Internatonal Conference on Informaton Scence, Sgnal Processng and ther pplcatons (ISSP 2012), 2 5 July, Montreal, Canada, pp Impedovo, D., Prlo, G. and Petrone, M. (2011) multresoluton mult-classfer system for speaker verfcaton, Expert Systems, Vol. 29, No. 5, pp Impedovo, D., Prlo, G. and Refce, M. (2008) Handwrtten sgnature and speech: prelmnary experments on multple source and classfers for personal dentty verfcaton, LNCS, Vol. 5158, pp Impedovo, D., Prlo, G., Sarcnella, L. and Stasolla, E. (2010) rtfcal classfer generaton for mult-expert system evaluaton, Proceedngs of the IEEE Internatonal Conference on Fronters n Handwrtng Recognton (ICFHR 2010), November, Kolkata, Inda, pp Kttler, J., Hatef, M., Dun, R.P.W. and Matas, J. (1998) On combnng classfers, IEEE Transactons on Pattern nalyss Machne Intellgence, Vol. 20, No.3, pp Lu, C.L., Nakashma, K., Sako, H. and Fujsawa, H. (2003) Handwrtten dgt recognton: benchmarkng of state-of-the-art technques, Pattern Recognton, Vol. 36, No. 10, pp Prlo, G. and Impedovo, D. (2011) Fuzzy-zonng-based classfcaton for handwrtten characters, IEEE Transactons on Fuzzy Systems, Vol. 19, No. 4, pp Prlo, G., Trullo, C.. and Impedovo, D. (2009) feedback-based mult-classfer system, Proceedngs of the IEEE Internatonal Conference on Document nalyss and Recognton (ICDR 2009), July, Barcelona, Span, pp Plamondon, R. and Srhar, S.N. (2000) On-lne and off-lne handwrtng recognton: a comprehensve survey, IEEE Transactons on Pattern nalyss and Machne Intellgence, Vol. 22, No. 1, pp Polkar, R. (2007) Bootstrap-nspred technques n computatonal ntellgence, IEEE Sgnal Processng Magazne, Vol. 24, No. 4, pp Rattan,., Fren, B., Marcals, G.L. and Rol, F. (2009) Template update methods n adaptve bometrc systems: a crtcal revew, Proceedngs of the 3rd Internatonal Conference on Bometrcs, 2 5 June, lghero, Italy, LNCS, Vol. 5558, pp Suen, C.Y. and Tan, J. (2005) nalyss of errors of handwrtten dgts made by a multtude of classfers, Pattern Recognton Letters, Vol. 26, No. 3, pp Suen, C.Y., Nadal, C., Legault, R., Ma, T.. and Lam, L. (1992) Computer recognton of unconstraned handwrtten numerals, Proceedngs of the IEEE, Vol. 80, No. 7, pp Uludag, U., Ross,. and Jan,. (2004) Bometrc template selecton and update: a case study n fngerprnts, Pattern Recognton, Vol. 37, No. 7, pp

Benchmarking of Update Learning Strategies on Digit Classifier Systems

Benchmarking of Update Learning Strategies on Digit Classifier Systems 2012 Internatonal Conference on Fronters n Handwrtng Recognton Benchmarkng of Update Learnng Strateges on Dgt Classfer Systems D. Barbuzz, D. Impedovo, G. Prlo Dpartmento d Informatca Unverstà degl Stud

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

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

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

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 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

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task Proceedngs of NTCIR-6 Workshop Meetng, May 15-18, 2007, Tokyo, Japan Term Weghtng Classfcaton System Usng the Ch-square Statstc for the Classfcaton Subtask at NTCIR-6 Patent Retreval Task Kotaro Hashmoto

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

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

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 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

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

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

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

Online Detection and Classification of Moving Objects Using Progressively Improving Detectors

Online Detection and Classification of Moving Objects Using Progressively Improving Detectors Onlne Detecton and Classfcaton of Movng Objects Usng Progressvely Improvng Detectors Omar Javed Saad Al Mubarak Shah Computer Vson Lab School of Computer Scence Unversty of Central Florda Orlando, FL 32816

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

Classifying Acoustic Transient Signals Using Artificial Intelligence

Classifying Acoustic Transient Signals Using Artificial Intelligence Classfyng Acoustc Transent Sgnals Usng Artfcal Intellgence Steve Sutton, Unversty of North Carolna At Wlmngton (suttons@charter.net) Greg Huff, Unversty of North Carolna At Wlmngton (jgh7476@uncwl.edu)

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

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

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

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

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

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

PRÉSENTATIONS DE PROJETS

PRÉSENTATIONS DE PROJETS PRÉSENTATIONS DE PROJETS Rex Onlne (V. Atanasu) What s Rex? Rex s an onlne browser for collectons of wrtten documents [1]. Asde ths core functon t has however many other applcatons that make t nterestng

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

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

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

Related-Mode Attacks on CTR Encryption Mode

Related-Mode Attacks on CTR Encryption Mode Internatonal Journal of Network Securty, Vol.4, No.3, PP.282 287, May 2007 282 Related-Mode Attacks on CTR Encrypton Mode Dayn Wang, Dongda Ln, and Wenlng Wu (Correspondng author: Dayn Wang) Key Laboratory

More information

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data Malaysan Journal of Mathematcal Scences 11(S) Aprl : 35 46 (2017) Specal Issue: The 2nd Internatonal Conference and Workshop on Mathematcal Analyss (ICWOMA 2016) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES

More information

An Entropy-Based Approach to Integrated Information Needs Assessment

An Entropy-Based Approach to Integrated Information Needs Assessment Dstrbuton Statement A: Approved for publc release; dstrbuton s unlmted. An Entropy-Based Approach to ntegrated nformaton Needs Assessment June 8, 2004 Wllam J. Farrell Lockheed Martn Advanced Technology

More information

Incremental MQDF Learning for Writer Adaptive Handwriting Recognition 1

Incremental MQDF Learning for Writer Adaptive Handwriting Recognition 1 200 2th Internatonal Conference on Fronters n Handwrtng Recognton Incremental MQDF Learnng for Wrter Adaptve Handwrtng Recognton Ka Dng, Lanwen Jn * School of Electronc and Informaton Engneerng, South

More information

UB at GeoCLEF Department of Geography Abstract

UB at GeoCLEF Department of Geography   Abstract UB at GeoCLEF 2006 Mguel E. Ruz (1), Stuart Shapro (2), June Abbas (1), Slva B. Southwck (1) and Davd Mark (3) State Unversty of New York at Buffalo (1) Department of Lbrary and Informaton Studes (2) Department

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 New Approach For the Ranking of Fuzzy Sets With Different Heights

A New Approach For the Ranking of Fuzzy Sets With Different Heights New pproach For the ankng of Fuzzy Sets Wth Dfferent Heghts Pushpnder Sngh School of Mathematcs Computer pplcatons Thapar Unversty, Patala-7 00 Inda pushpndersnl@gmalcom STCT ankng of fuzzy sets plays

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

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

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

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

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

Optimizing Document Scoring for Query Retrieval

Optimizing Document Scoring for Query Retrieval Optmzng Document Scorng for Query Retreval Brent Ellwen baellwe@cs.stanford.edu Abstract The goal of ths project was to automate the process of tunng a document query engne. Specfcally, I used machne learnng

More information

An Approach to Real-Time Recognition of Chinese Handwritten Sentences

An Approach to Real-Time Recognition of Chinese Handwritten Sentences An Approach to Real-Tme Recognton of Chnese Handwrtten Sentences Da-Han Wang, Cheng-Ln Lu Natonal Laboratory of Pattern Recognton, Insttute of Automaton of Chnese Academy of Scences, Bejng 100190, P.R.

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

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

Recognition of Handwritten Numerals Using a Combined Classifier with Hybrid Features

Recognition of Handwritten Numerals Using a Combined Classifier with Hybrid Features Recognton of Handwrtten Numerals Usng a Combned Classfer wth Hybrd Features Kyoung Mn Km 1,4, Joong Jo Park 2, Young G Song 3, In Cheol Km 1, and Chng Y. Suen 1 1 Centre for Pattern Recognton and Machne

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

A Lazy Ensemble Learning Method to Classification

A Lazy Ensemble Learning Method to Classification IJCSI Internatonal Journal of Computer Scence Issues, Vol. 7, Issue 5, September 2010 ISSN (Onlne): 1694-0814 344 A Lazy Ensemble Learnng Method to Classfcaton Haleh Homayoun 1, Sattar Hashem 2 and Al

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

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

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

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

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

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

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

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

Manifold-Ranking Based Keyword Propagation for Image Retrieval *

Manifold-Ranking Based Keyword Propagation for Image Retrieval * Manfold-Rankng Based Keyword Propagaton for Image Retreval * Hanghang Tong,, Jngru He,, Mngjng L 2, We-Yng Ma 2, Hong-Jang Zhang 2 and Changshu Zhang 3,3 Department of Automaton, Tsnghua Unversty, Bejng

More information

Reducing Frame Rate for Object Tracking

Reducing Frame Rate for Object Tracking Reducng Frame Rate for Object Trackng Pavel Korshunov 1 and We Tsang Oo 2 1 Natonal Unversty of Sngapore, Sngapore 11977, pavelkor@comp.nus.edu.sg 2 Natonal Unversty of Sngapore, Sngapore 11977, oowt@comp.nus.edu.sg

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

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

Foreground and Background Information in an HMM-based Method for Recognition of Isolated Characters and Numeral Strings

Foreground and Background Information in an HMM-based Method for Recognition of Isolated Characters and Numeral Strings Foreground and Background Informaton n an HMM-based Method for Recognton of Isolated Characters and Numeral Strngs Alceu de S. Brtto Jr a,b, Robert Sabourn c,d, Flavo Bortolozz a,chng Y. Suen d a Pontfíca

More information

y and the total sum of

y and the total sum of Lnear regresson Testng for non-lnearty In analytcal chemstry, lnear regresson s commonly used n the constructon of calbraton functons requred for analytcal technques such as gas chromatography, atomc absorpton

More information

Spam Filtering Based on Support Vector Machines with Taguchi Method for Parameter Selection

Spam Filtering Based on Support Vector Machines with Taguchi Method for Parameter Selection E-mal Spam Flterng Based on Support Vector Machnes wth Taguch Method for Parameter Selecton We-Chh Hsu, Tsan-Yng Yu E-mal Spam Flterng Based on Support Vector Machnes wth Taguch Method for Parameter Selecton

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

The Research of Ellipse Parameter Fitting Algorithm of Ultrasonic Imaging Logging in the Casing Hole

The Research of Ellipse Parameter Fitting Algorithm of Ultrasonic Imaging Logging in the Casing Hole Appled Mathematcs, 04, 5, 37-3 Publshed Onlne May 04 n ScRes. http://www.scrp.org/journal/am http://dx.do.org/0.436/am.04.584 The Research of Ellpse Parameter Fttng Algorthm of Ultrasonc Imagng Loggng

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

Accumulated-Recognition-Rate Normalization for Combining Multiple On/Off-Line Japanese Character Classifiers Tested on a Large Database

Accumulated-Recognition-Rate Normalization for Combining Multiple On/Off-Line Japanese Character Classifiers Tested on a Large Database 4 th Internatonal Workshop on Multple Classfer Systems (MCS23) Guldford, UK Accumulated-Recognton-Rate Normalzaton for Combnng Multple On/Off-Lne Japanese Character Classfers Tested on a Large Database

More information

Some Advanced SPC Tools 1. Cumulative Sum Control (Cusum) Chart For the data shown in Table 9-1, the x chart can be generated.

Some Advanced SPC Tools 1. Cumulative Sum Control (Cusum) Chart For the data shown in Table 9-1, the x chart can be generated. Some Advanced SP Tools 1. umulatve Sum ontrol (usum) hart For the data shown n Table 9-1, the x chart can be generated. However, the shft taken place at sample #21 s not apparent. 92 For ths set samples,

More information

Complex Numbers. Now we also saw that if a and b were both positive then ab = a b. For a second let s forget that restriction and do the following.

Complex Numbers. Now we also saw that if a and b were both positive then ab = a b. For a second let s forget that restriction and do the following. Complex Numbers The last topc n ths secton s not really related to most of what we ve done n ths chapter, although t s somewhat related to the radcals secton as we wll see. We also won t need the materal

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

Complex System Reliability Evaluation using Support Vector Machine for Incomplete Data-set

Complex System Reliability Evaluation using Support Vector Machine for Incomplete Data-set Internatonal Journal of Performablty Engneerng, Vol. 7, No. 1, January 2010, pp.32-42. RAMS Consultants Prnted n Inda Complex System Relablty Evaluaton usng Support Vector Machne for Incomplete Data-set

More information

Learning-based License Plate Detection on Edge Features

Learning-based License Plate Detection on Edge Features Learnng-based Lcense Plate Detecton on Edge Features Wng Teng Ho, Woo Hen Yap, Yong Haur Tay Computer Vson and Intellgent Systems (CVIS) Group Unverst Tunku Abdul Rahman, Malaysa wngteng_h@yahoo.com, woohen@yahoo.com,

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

F Geometric Mean Graphs

F Geometric Mean Graphs Avalable at http://pvamu.edu/aam Appl. Appl. Math. ISSN: 1932-9466 Vol. 10, Issue 2 (December 2015), pp. 937-952 Applcatons and Appled Mathematcs: An Internatonal Journal (AAM) F Geometrc Mean Graphs A.

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

Load Balancing for Hex-Cell Interconnection Network

Load Balancing for Hex-Cell Interconnection Network Int. J. Communcatons, Network and System Scences,,, - Publshed Onlne Aprl n ScRes. http://www.scrp.org/journal/jcns http://dx.do.org/./jcns.. Load Balancng for Hex-Cell Interconnecton Network Saher Manaseer,

More information

A Weighted Method to Improve the Centroid-based Classifier

A Weighted Method to Improve the Centroid-based Classifier 016 Internatonal onference on Electrcal Engneerng and utomaton (IEE 016) ISN: 978-1-60595-407-3 Weghted ethod to Improve the entrod-based lassfer huan LIU, Wen-yong WNG *, Guang-hu TU, Nan-nan LIU 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

A Background Subtraction for a Vision-based User Interface *

A Background Subtraction for a Vision-based User Interface * A Background Subtracton for a Vson-based User Interface * Dongpyo Hong and Woontack Woo KJIST U-VR Lab. {dhon wwoo}@kjst.ac.kr Abstract In ths paper, we propose a robust and effcent background subtracton

More information

TN348: Openlab Module - Colocalization

TN348: Openlab Module - Colocalization TN348: Openlab Module - Colocalzaton Topc The Colocalzaton module provdes the faclty to vsualze and quantfy colocalzaton between pars of mages. The Colocalzaton wndow contans a prevew of the two mages

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

Data Mining: Model Evaluation

Data Mining: Model Evaluation Data Mnng: Model Evaluaton Aprl 16, 2013 1 Issues: Evaluatng Classfcaton Methods Accurac classfer accurac: predctng class label predctor accurac: guessng value of predcted attrbutes Speed tme to construct

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

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

SVM-based Learning for Multiple Model Estimation

SVM-based Learning for Multiple Model Estimation SVM-based Learnng for Multple Model Estmaton Vladmr Cherkassky and Yunqan Ma Department of Electrcal and Computer Engneerng Unversty of Mnnesota Mnneapols, MN 55455 {cherkass,myq}@ece.umn.edu Abstract:

More information

Fingerprint matching based on weighting method and SVM

Fingerprint matching based on weighting method and SVM Fngerprnt matchng based on weghtng method and SVM Ja Ja, Lanhong Ca, Pnyan Lu, Xuhu Lu Key Laboratory of Pervasve Computng (Tsnghua Unversty), Mnstry of Educaton Bejng 100084, P.R.Chna {jaja}@mals.tsnghua.edu.cn

More information

Keywords - Wep page classification; bag of words model; topic model; hierarchical classification; Support Vector Machines

Keywords - Wep page classification; bag of words model; topic model; hierarchical classification; Support Vector Machines (IJCSIS) Internatonal Journal of Computer Scence and Informaton Securty, Herarchcal Web Page Classfcaton Based on a Topc Model and Neghborng Pages Integraton Wongkot Srura Phayung Meesad Choochart Haruechayasak

More information

Pruning Training Corpus to Speedup Text Classification 1

Pruning Training Corpus to Speedup Text Classification 1 Prunng Tranng Corpus to Speedup Text Classfcaton Jhong Guan and Shugeng Zhou School of Computer Scence, Wuhan Unversty, Wuhan, 430079, Chna hguan@wtusm.edu.cn State Key Lab of Software Engneerng, Wuhan

More information

Investigating the Performance of Naïve- Bayes Classifiers and K- Nearest Neighbor Classifiers

Investigating the Performance of Naïve- Bayes Classifiers and K- Nearest Neighbor Classifiers Journal of Convergence Informaton Technology Volume 5, Number 2, Aprl 2010 Investgatng the Performance of Naïve- Bayes Classfers and K- Nearest Neghbor Classfers Mohammed J. Islam *, Q. M. Jonathan Wu,

More information

Journal of Engineering Science and Technology Review 7 (3) (2014) Research Article

Journal of Engineering Science and Technology Review 7 (3) (2014) Research Article Jestr Journal of Engneerng cence and Technology Revew 7 (3) (2014) 151 157 Research Artcle JOURAL OF Engneerng cence and Technology Revew www.estr.org Traffc Classfcaton Method by Combnaton of Host Behavour

More information

Conditional Speculative Decimal Addition*

Conditional Speculative Decimal Addition* Condtonal Speculatve Decmal Addton Alvaro Vazquez and Elsardo Antelo Dep. of Electronc and Computer Engneerng Unv. of Santago de Compostela, Span Ths work was supported n part by Xunta de Galca under grant

More information

Analysis of Continuous Beams in General

Analysis of Continuous Beams in General Analyss of Contnuous Beams n General Contnuous beams consdered here are prsmatc, rgdly connected to each beam segment and supported at varous ponts along the beam. onts are selected at ponts of support,

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

Cordial and 3-Equitable Labeling for Some Star Related Graphs

Cordial and 3-Equitable Labeling for Some Star Related Graphs Internatonal Mathematcal Forum, 4, 009, no. 31, 1543-1553 Cordal and 3-Equtable Labelng for Some Star Related Graphs S. K. Vadya Department of Mathematcs, Saurashtra Unversty Rajkot - 360005, Gujarat,

More information

Face Recognition Based on SVM and 2DPCA

Face Recognition Based on SVM and 2DPCA Vol. 4, o. 3, September, 2011 Face Recognton Based on SVM and 2DPCA Tha Hoang Le, Len Bu Faculty of Informaton Technology, HCMC Unversty of Scence Faculty of Informaton Scences and Engneerng, Unversty

More information

Sixth Indian Conference on Computer Vision, Graphics & Image Processing

Sixth Indian Conference on Computer Vision, Graphics & Image Processing Sxth Indan Conference on Computer Vson, Graphcs & Image Processng Incorporatng Cohort Informaton for Relable Palmprnt Authentcaton Ajay Kumar Bometrcs Research Laboratory, Department of Electrcal Engneerng

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

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

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

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

Title: A Novel Protocol for Accuracy Assessment in Classification of Very High Resolution Images

Title: A Novel Protocol for Accuracy Assessment in Classification of Very High Resolution Images 2009 IEEE. Personal use of ths materal s permtted. Permsson from IEEE must be obtaned for all other uses, n any current or future meda, ncludng reprntng/republshng ths materal for advertsng or promotonal

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