An Item-Targeted User Similarity Method for Data Service Recommendation

Size: px
Start display at page:

Download "An Item-Targeted User Similarity Method for Data Service Recommendation"

Transcription

1 01 IEEE 16th Internatonal Enterprse Dstrbuted Object Computng Conference Workshops An Item-Targeted User Smlarty Method for Data Serce Recommendaton Cheng Zhang, Xaofang Zhao Insttute of Computng Technology Chnese Academy of Scences Bejng, Chna Janwu Wang San Dego Supercomputer Center Unersty of Calforna, San Dego San Dego, U.S.A Abstract Memory-based methods for recommendng data serces predct the ratngs of acte users based on the nformaton of other smlar users or tems, where the smlarty algorthm always plays a key role. In many scenaros, we fnd that the smlarty of two users always show dfferent effecteness when predctng dfferent ratngs. Normal smlarty algorthms usually do not count the dfference, snce they orgnate from statstc and algebra felds and do not drectly am at recommendatons. Ths paper proposes a noel method to amend the user smlarty generated by a normal smlarty algorthm to more accurately descrbe the effecteness of the smlarty on a targeted tem. We apply our method to mproe the Pearson Correlaton Coeffcent (PCC) algorthm whch s one of the most commonly used smlarty algorthms. The experment results on some practcal datasets show that our method s slghtly better than the orgnal PCC algorthm for predctng ratngs n recommendatons. Keywords-data serces; recommender system; user smlarty; collaborate flterng I. INTRODUCTION Data serce recommendaton predcts the nterest of data serces(tems) for an acte user by collectng ratng nformaton from other smlar users or tems, and related technques hae been wdely employed n some famous web communtes [14], such as Amazon 1 and Ebay. The key dea s that the acte user wll prefer those tems that are preferred by smlar users. Memory-based approaches hae been deeply studed and wdely used n practce, whch can be dded n two categores: user-based [4, 8, 10] and tem-based [17, 6]. When estmatng an unknown test ratng, memory-based approaches wll frstly collect the smlar users or tems by some smlarty algorthms, such as Pearson Correlaton Coeffcent (PCC) [16] and Vector Space Smlarty (VSS) [4]. Then, the ratng nformaton of smlar users or tems wll be handled by some predcton algorthms to produce the estmated alue for the test ratng. In many scenaros, we fnd that the smlarty of two users always show dfferent effecteness when predctng dfferent ratngs. For example, user u and are smlar users snce they share the preference on tems 1,, 3, 4, whch are all data serces on comedy moes. Item 5 and 6 are unknown ratngs of user where 5 s a data serce on comedy moes and 6 s a data serce on scence-fcton moe. The ratngs of user u on tem 5 and 6 wll be used to predct the ratng of 5 and 6 for user respectely accordng to the smlarty between u and. The smlarty wll be more effecte to predct the ratng of user on tem 5 than that on tem 6, due to our common experences that t s unreasonable to nfer a user s preference on a scencefcton moe from hs/her preference on comedy moes. Smlarly, f user w and are smlar because they both lke seeral acton moes, the smlarty between user u and wll be more effecte than the smlarty between w and to predct the ratng of user on tem 5. Yet normal smlarty algorthms usually do not count such dfferences, snce they generate from more unersal statstc prncples and do not drectly am at recommendatons. So, there s a latent error factor when applyng such algorthms nto the recommendaton tasks drectly. Based on the aboe obseratons, we argue that the aspect of the predcted tem should be elaborately counted n the user smlarty calculaton to get better recommendatons. Ths paper proposes a noel method to amend the user smlarty generated by a normal smlarty algorthm to descrbe the effecteness of the smlarty on a targeted tem more accurately. Gen a specfc smlarty algorthm, our method can work n two ways: one s called Correcton Polcy, whch uses tem relatonshps to adjust the user smlarty; the other s named Precson Polcy, whch selects the more sutable co-rated tems to calculate user smlartes. We note that ths paper focuses on the effecteness descrpton of the user smlarty on dfferent tems and the nfluence analyses of the effecteness on the ratng predcton, rather than addressng a combned model to get the optmal recommendaton as we hae learnt from the Netflx Prze competton [3]. So, we apply our method to mproe the Pearson Correlaton Coeffcent (PCC) algorthm whch s one of the most commonly used smlarty algorthms. The experment results on some practcal datasets show that our method outperforms the orgnal PCC algorthm for predctng ratngs n recommendatons. The rest of the paper s organzed as follows. The next secton prodes a bref reew of related work. Secton 3 presents our method of the tem-targeted user smlarty. The /1 $ IEEE DOI /EDOCW

2 results of an emprcal analyss are presented n Secton 4, followed by a concluson n Secton 5. II. RELATED WORK A. Defntons and Notatons Let I { 1,,..., n } be a set of tems, U {u 1, u,..., u m } be a set of users, and I u be the set of tems rated by u. The relatonshp between users and tems s denoted by an MN matrx, where each entry r represents the ratng of tem made by user u. Let r u represent the aerage rate of user u on all rated tems. B. Recommendaton Methods We wll frst brefly reew the deelopment n recommendaton. Generally, exsted algorthms fall n ether of the followng two categores: memory-based flterng and model-based flterng [7]. 1)Memory-based Approaches.Memory-based approaches essentally are heurstcs to predct a mssng ratng by aggregatng the ratng of k-neghbors who hae preously rated the tem. To dentfy the k-neghbors, seeral algorthms hae been proposed to compute the smlarty between each par of users or tems. The common-used aggregaton s to compute the weghted aerage of the k- neghbors ratngs on the partcular tem [1]. Consderng the fact that dfferent users may use the ratng scale dfferently, the weghted sum uses ther deatons from the aerage ratng of the correspondng neghbor. Two types of memorybased approaches hae been studed: user-based [4, 8,10] and tem-based [17, 6]. User-based approaches predct the ratngs of acte users based on the ratngs of smlar users found, and tem-based approaches predct the ratngs based on the nformaton of the computed smlar tems. )Model-based Approaches. In contrast to memorybased methods, model-based approaches use the tranng examples to learn a predefned model, whch s then used to make ratng predctons. There are arous approaches n ths category. Typcal examples nclude clusterng model [1, 0, 11], aspect models [9, 19] and latent factor models [5]. Dmensonalty reducton and matrx reconstructon technques are often the subject of dscusson [, 18]. Model-based approaches are often tme-consumng to buld and update, and cannot coer as derse a user range as the memory-based approaches do [3]. In order to take the adantages of memory-based approaches and model-based approaches, hybrd approaches hae been studed recently. For example, Bell et al. combned the output of multple recommender algorthms to mproe performance [1, ]. 3)Smlarty Computatons. As mentoned aboe, smlarty algorthms are used to collect the smlar neghbors n memory-based approaches, and drectly affect predctons accuracy. Varous approaches hae been used to compute the smlarty between users or tems n recommender systems. We wll take user smlartes as examples to explan the prncpal of these approaches. The two most popular approaches are Pearson Correlaton Coeffcent (PCC) [16] and Vector Space Smlarty (VSS) [4]. PCC algorthm [16] measures the smlarty between two users based on the tems that both users hae rated n common: Sm( ) I I r ) u r ) u I r ) r ) where Sm( ) denotes the smlarty between user u and user and belongs to the subset of tems co-rated by both users u and. Obously, Sm( ) s rangng from [0, 1]. VSS algorthm [4] measures the smlarty of user u and user by computng the cosne angles between the ratng ectors of them: I u I ru, I I (1) ru, r, () Sm( ) cos( ) r where u and denotes the ratng ectors of user u and respectely. Generally, PCC approach s more accurate than VSS approach, snce t consders the factor of the dfferences of user ratng styles [14]. The adjusted cosne smlarty compensates ths drawback by subtractng the correspondng user aerage from each co-rated par and has been shown more effecte for measurng tem smlartes [17]. On the other hand, PCC algorthm oerestmates the smlartes of users who happen to hae rated a few tems dentcally, but may not hae smlar oerall preferences [15]. Herlocker etc. proposed to use the followng equaton to modfy the user smlarty [15]. Max( Iu I Sm' ( ) γ, γ ) Sm( ) Here s a threshold parameter. Ths method soled the problem when only few tems rated but n case that when I a I b s much hgher than, the modfed smlarty wll be larger than 1. In order to bound the smlarty alue to the nteral [0,1], Hao etc. proposed another equaton [14]: Mn( Iu I Sm' ( ) γ (3), γ ) Sm( ) (4) Some other researchers address the fact that the smlarty algorthms would not work well when there are few tems co-rated by both user u and. It was emprcally shown that the ratng predcton accuracy could mproe f we frstly smooth the mssng data of the user-tem matrx [14, 1]. III. ITEM-TARGETED USER SIMILARITY The basc dea of the tem-targeted user smlarty s to amend the user smlarty generated by a normal smlarty algorthm to descrbe the effecteness of the smlarty on a targeted tem more accurately. We wll explan our method 173

3 based on the PCC algorthm snce t s one of the most commonly used smlarty algorthms. Our method can work n two ways to calculate the tem-targeted user smlarty. In the followng two sub-sectons, we wll dscuss them n detals. A. Correcton Polcy The key dea of Correcton Polcy s to adjust the orgnal user smlarty usng a correcton factor whch measures the smlarty effecteness on the predcted ratng, and the orgnal user smlarty s calculated by a normal smlarty algorthm, such as PCC. The correcton factor s as follows: SmQau Sm( j, j IU I ( (5) Iu I where s the predcted ratng. Gen a specfc tem, the aboe equaton ealuates the qualty of the user smlarty accordng to the smlarty relatonshps between the tem and the tems whch are shared by the user u and. We call ths factor as Smlarty Qualty. By defnton, the alue of Smlarty Qualty les n the range [0,1]. Also, the more smlar the shared tems and the tem are, the larger the SmQau alue s. Based on Smlarty Qualty, the resed smlarty between two users for a specfc tem s as follows: Sm( Sm( ) SmQau( (6) Here Sm( ) les n the nteral [-1, 1]. Sm( wll be an order of magntude lower than Sm( ) snce the alue of SmQau( les n the nteral [0, 1]. So the contrbuton of smlar users on the predcted ratng wll be largely reduced. It s a reasonable nference f we predct the unknown ratng usng the resed smlarty, the predcted ratng wll be less than t should be. Therefore, a recoery functon s requred to ensure the resed smlarty can keep the effecteness of the smlarty qualty whle remong the harmful mpact mentoned aboe. So that the smlarty s further resed as: Sm( F( Sm( ) SmQau( ) x F ( x), x[-1,1] (7) x Obously, when the alue of x les n [-1, 1], the nteral of F(x) s stll n [-1,1] and meanwhle F(x) s larger than the absolute alue of x. In Secton 4, we wll test the effect of the method on some practcal datasets. B. Precson Polcy The key dea of Precson Polcy s to select such co-rated tems that are smlar to the specfc tem to calculate the smlarty between user u and. Based on the PCC algorthm, we measure the tem-targeted smlarty between two users by addng another factor : Sm( j I (, ) >, )(, ) I Sm j u j ru r j r u α (8) r ) r ) j Where s the tem smlarty threshold, and Sm(, j) s computed by the normal PCC algorthm. The selecton of the parameter s mportant snce a bg alue wll always cause the shortage of smlar users, and a relate small alue wll not get rd of the mpact of the dssmlar tems. In practce, two smlar users do not always share such tems that are hghly smlar to the predcted tem. In order to adequately utlze the nformaton of such users, the temtargeted smlarty s calculated as the follow steps: Step 1: To compute the smlarty of two users by the Eq.1; Step : If the smlarty s less than, then t s recalculate by the Eq.8 Where s a alue assgned beforehand. The adantage of ths method s that we can dscoer the smlarty between users as much as possble, whch wll prode more useful nformaton for the predcton task. Ths combned method wll be referred to as c-precson Polcy and the raw precson polcy expressed by the Eq.8 s called r-precson Polcy. In Secton 4, we wll tune the parameter based on our experments and present the emprcal result. C. Recommendaton Framework Wth the aboe tem-targeted smlarty algorthm, there are stll seeral basc problems for recommendatons. One s how to select smlar users, and another s how to estmate the unknown ratng usng the smlar users nformaton. The frst queston comes from the fact that the smlar users wth a lower smlarty alue wll decrease the accuracy of the predcton. Although the tem-targeted smlarty may hae an adantage oer the orgnal smlarty, we stll cannot neglect ths problem. In ths paper, we adopt the method proposed n the paper [14] to deal wth ths problem. Before makng a predcton, a set of k most smlar users S(u) toward user u are frst generated accordng to: u j { u Sm( u, > u u} S( u) θ, (9) a a where Sm(u a, s calculated usng Eq.7 or Eq.8 respectely correspondng to the Correcton Polcy and the Precson Polcy, and s the smlarty threshold. Accordng to Eq.9, the canddate user wll be selected as the smlar user f the smlarty between the canddate user and the acte user s larger than, whch tells us that the selecton of s mportant. We wll tune the parameter based on our experments and present the emprcal results n Secton 4. To sole the second queston, a sum of the aerage ratng made by the acte user u and a weghted aerage of the smlar users ratngs on the specfc tem s used to generate the predcton on the unknown ratng: a 174

4 S( u) r ru + (10) S ( u) r ) Sm( Sm( where r _ u denotes the aerage ratng made by user u and Sm(, whch s computed by Eq.7 or Eq.8, s the tem-targeted smlarty between user u and user. Ths weghted sum of deatons from the smlar users aerage ratng specally take nto account the fact that dfferent users may use ratng scale dfferently. IV. EXPERIMENTS We dd multple experments to ealuate the effecteness of our method. In partcular, we address the followng ssues: How do the recoery functon F and the threshold affect the accuracy of predctons n the correcton polcy? How do the thresholds and affect the accuracy of predctons n the precson polcy? For the two tem-targeted smlarty polces, namely Correcton Polcy and Precson Polcy (ncludng c- precson polcy and r-precson polcy), whch one more effecte? Is the tem-targeted user smlarty better for predctons compared wth the user smlarty whch our method amends? A. Datasets Two datasets from moe ratng are used n our experments: MoeLens 3 and Netflx 4. We wll report the smulaton results n detals. TheMoeLens dataset contans 100,000 ratngs (1-5 scales) rated by 943 users on 168 moes, and each user at least rated 0 moes. The densty of ths datasets s 8.77%. To test on dfferent numbers of tranng users, we extracted a subset of 500 users wth more than 40 ratngs. The frst 00 users are used for tranng and we altered the tranng sze to be 50, 100, 00 users. The last 300 users are used for test, and each ratng of these users was predcted usng an all-butone polcy [4]. The Netflx dataset contans oer 100 mllon ratngs (1-5 scales) rated by users on moes. We randomly extracted a subset of 500 users wth more than 80 ratngs on 3000 moes. The frst 00 users are used for tranng and we altered the tranng sze to be 100, 00 users. The last 300 users are used for test, and each ratng of these users was predcted usng an all-but-one polcy. B. Metrcs The major crteron for ealuatng ratng-orented recommendaton algorthms s the ratng predcton accuracy [13]. Seeral technques hae been proposed to achee t ncludng Mean Absolute Error (MAE), Mean Squared Error (MSE), Normalzed Mean Absolute Error (NMAE), and so on. In ths paper, the Mean Absolute Error metrcs s selected snce t works well for measurng how accurately the algorthm predcts the ratng of a randomly selected tem [10]. MAE s defned as: MAE r r u u N, ˆ, (11) where r denotes the ratng that user u gae to tem, and r ˆu, denotes the ratng that user u gae to tem whch s predcted by our approach, and N denotes the number of tested ratngs. C. Impact of Recoery Functon and Fgure 1. Impact of parameter F and on MoeLens dataset. Fgure. Impact of parameter F and on Netflx dataset. As shown n Eq.9, when predctng the ratngs under the correcton polcy, we ge a threshold to decde the selecton of smlar users whch has been collected. By assgnng a dfferent alue to, the performance of the predcton could be affected. The alue of s ared from 0 to 1. When settng as 0, the algorthm uses all the smlar users to complete the predcton. When s set as 1, only users that are equal to the acte user wll be used to predct the unknown ratng. We tune the alue to show the performance on predcton. Fg.1 and Fg. show the performance changes when the threshold ncreases from 0 to 1. We can see that the 175

5 correcton polcy wll achee the best performance around 0. on both the MoeLens dataset and the Netflx dataset. The change trends are smlar for dfferent numbers of users. D. Impact of Parameter When we adopt the r-precson polcy to mplement recommendatons, parameter may affect the accuracy of the tem-targeted smlarty between two users snce t s used to decde whch tems wll be selected for computng the smlarty alue. When settng as 0, all shared tems between two users wll be used to compute the smlarty, and so the tem-targeted smlarty algorthm s equal to the normal PCC algorthm. In ths experment, we ared the range of from 0 to 1 wth a step alue of 0.1. Fg.3 and Fg.4 show how the parameter affects the accurcy of predctons on the MoeLens dataset and the Netflx dataset respectely. enables the algorthm to aod the recalculaton for the users wth a hgh smlarty alue. Based on the aboe tranng results, we tested the parameter on the MoeLens and the Netflx datasets. For the MoeLens dataset, we assgned 0.4 to and selected 50, 100 and 00 tranng users. For the Netflx dataset, we assgned 0.8 to and selected 100 and 00 tranng users. In ths experment, we ared the alue of from 0 to 1 wth a step alue of 0.1. Fg.5 and Fg.6 tell us how the parameter affects the accuracy of predctons based on the all-but-one polcy Fgure 5. Impact of parameter on MoeLens dataset Fgure 3. Impact of parameter F and on Netflx dataset. Fgure 6. Impact of parameter on Netflx dataset Fgure 4. Impact of parameter F and on Netflx dataset. From Fg.3, we can see that the r-precson polcy achees the best performance around 0.4 on the MoeLens dateset. Fg.4 shows that the r-precson polcy achees the best performance around 0.8 on the Netflx dateset. Although the optmal alue of s dfferent on the two datasets, the change trends are smlar for dfferent numbers of users on a certan dataset. E. Impact of Parameter We expect that parameter can make predctons based on the c-precson polcy more accurate n practce snce t As showed n Fg.5, gen 50, 100 and 00 tranng users respectely, ths c-precson polcy wll achee the best performance around 0.1 on the MoeLens dataset. Fg.6 shows that ths polcy wll achee the best performance around 0 on the Netflx dataset when selectng 100 and 00 tranng users respectely. Also, the changng tendences are also smlar at dfferent user numbers. Ths result tells us that the c-precson polcy hae no obous effects on mprong the predcton accuracy. F. Comparsons We note that ths paper focuses on descrbng the effecteness of the user smlarty on dfferent tems and analyzng the nfluences of the effecteness on the ratng predcton, rather than addressng a combned model to get the optmal recommendaton as we hae learnt from the Netflx Prze competton [3]. We dscuss how to mplement our method by amendng a normal algorthm, whch wll be called orgnal algorthm herenafter. So, we wll compare 176

6 our method wth the orgnal algorthm whch s the PCC algorthm n ths paper. Snce the c-precson polcy s a weak method for mprong the predcton accuracy, we manly compare the orgnal PCC algorthm wth our Correcton Polcy and r- Precson Polcy n ths experment. The ealuaton on the MoeLens and the Netflx datasets s done based on allbut-one polcy. For the MoeLens dataset, we set 0.4, 0. as suggested by the aboe results and ary the number of test users from 50 to 00. For the Netflx dataset, we set 0.8, 0. and ary the number of test users from 100 to 300. Note that the test users and the tranng users come from dfferent user sets and the detals hae been explaned n Secton 4.1. The results are shown n Tab.1 and Tab. respectely. TABLE I. MAE ON MOVIELENS DATASET FOR THE COMPARED ALGORITHMS 50 users 100 users 00 users PCC algorthm Correcton Polcy r-precson Polcy TABLE II. MAE ON NETFLIX DATASET FOR THE COMPARED ALGORITHMS 100 users 00 users 300 users PCC algorthm Correcton Polcy r-precson Polcy We note that the performance of dfferent algorthms would mproe along wth the ncrease of the test user numbers. Ths s because, wth a larger user database, t wll be easer to fnd suffcent number of neghbors wth hgh smlartes to the acte user so that hs preferences can be estmated more accurately. Also, our method ges better results on the Netflx dataset, snce ths dataset hae greater number of moes and t wll be easer to fnd more smlar tems to calculate the tem-targeted smlarty. The aboe test results show that the tem-targeted smlarty can nduce more accurate precsons than the orgnal algorthm. V. CONCLUSIONS In ths paper, we propose an tem-targeted smlarty method for data serce recommendaton based on the obseraton that the effecteness of the user smlarty s subject to the predcted tem. Our method takes nto account the relatonshp of the predcted tem and the tems used by smlarty computatons. We select the PCC algorthm as the orgnal algorthm and dese two polces for computng tem-targeted smlartes: one s the correcton polcy and the other s the precson polcy. The experment results on some practcal datasets show that our method outperforms the orgnal smlarty method when predctng ratngs n recommendatons. For future work, we plan to do more research on how to apply our methodology n the calculaton of tem smlartes. We wll also try to combne our method wth some other technques and study ther usefulness to recommendatons ACKNOWLEDGMENT Ths work s supported n part by the Natonal Key Technology Research and Deelopment Program of the Mnstry of Scence and Technology of Chna under grants 01BAH19B00. REFERENCES [1] G. Adomacus and A. Tuzhln. Toward the next generaton of recommender systems: A surey of the state-of-the-art and possble extensons. IEEE Trans. on Knowl. and Data Eng., ol. 17(6), pp , 005. [] R. Bell, Y. Koren, and C. Volnsky. Modelng relatonshps at multple scales to mproe accuracy of large recommender systems. In Proceedngs of the 13th ACM SIGKDD nternatonal conference on Knowledge dscoery and data mnng (SIGKDD 07), pp , 007. [3] R. M. Bell and Y. Koren. Lessons from the Netflx prze challenge. SIGKDD Explor. Newsl., ol. 9(), pp , 007. [4] J. S. Breese, D. Heckerman, and C. Kade. Emprcalanalyss of predcte algorthms for collaborate flterng. Proc. the 14th Conference on Uncertanty n Artfcal Intellgence (UAI 98), pp.43 5, [5] J. Canny. Collaborate flterng wth pracy a factor analyss. Proc. the 5th annual nternatonal ACM SIGIR conference on Research and deelopment n nformaton retreal (SIGIR05), pp , 00. [6] M. Deshpande and G. Karyps. Item-based top-n recommendaton algorthms. ACM Trans. Inf. Syst., ol. (1), pp , 004. [7] D. Goldberg, D. Nchols, B. M. Ok, and D. Terry. Usng collaborate flterng to weae an nformaton tapestry. Commun. ACM, ol. 35(1), pp , December 199. [8] J. L. Herlocker, J. A. Konstan, A. Borchers, and J. Redl. An algorthmc framework for performng collaborate flterng. Proc. the nd annual nternatonal ACM SIGIR conference on Research and deelopment n nformaton retreal (SIGIR 99), pp , [9] T. Hofmann. Collaborate flterng a Gaussan probablstc latent semantc analyss. Proc. the 6th annual nternatonal ACM SIGIR conference on Research and deelopment n nformaon retreal (SIGIR 03), pp , 003. [10] R. Jn, J. Y. Cha, and L. S. An automatc weghtng scheme for collaborate flterng. Proc. the 7th annual nternatonal ACM SIGIR conference on Research and deelopment n nformaton retreal (SIGIR 07), pp , 004. [11] A. Kohrs and B. Meraldo. Clusterng for collaborate flterng applcatons. Computatonal Intellgence for Modellng, Control and Automaton. IOS, [1] Y. Koren. Factorzaton meets the neghborhood: a multfaceted collaborate flterng model. Proc. the 14th ACM SIGKDD nternatonal conference on Knowledge dscoery and data mnng (SIGKDD 08), pp , 008. [13] N. N. Lu and Q. Yang. Egenrank: A rankng-orented approach to collaborate flterng. Proc. the 31st Annual Internatonal ACM SIGIR Conference on Research and Deelopment n Informaton Retreal (SIGIR08), pp ,

7 [14] H. Ma, I. Kng, and M. R. Lyu. Effecte mssng data predcton for collaborate flterng. Proc. the 30th annual nternatonal ACM SIGIR conference on Research and deelopment n nformaton retreal (SIGIR07), pp , 007. [15] M. R. McLaughln and J. L. Herlocker. A collaborate flterng algorthm and ealuaton metrc that accurately model the user experence. In Proceedngs of the 7th annual nternatonal ACM SIGIR conference on Research and deelopment n nformaton retreal (SIGIR04), pp , 004. [16] P. Resnck, N. Iacoo M. Suchak, P. Bergstrom, and J. Redl. Grouplens: An open archtecture for collaborate flterng of netnews. Proc. ACM 1994 Conference on Computer Supported Cooperate Work (CSCW94), pp , [17] B. Sarwar, G. Karyps, J. Konstan, and J. Redl. Item-based collaborate flterng recommendaton algorthms. Proc. the 10th nternatonal conference on World Wde Web (WWW01), pp , 001. [18] B. M. Sarwar, G. Karyps, J. A. Konstan, and J. T. Redl. Applcaton of dmensonalty reducton n recommender system - a case study. In ACM WebKDD Workshop, 000. [19] L. S and R. Jn. Flexble mxture model for collaborate flterng. Proc. the Twenteth Internatonal Conference on Machne Learnng (ICML 03), pp , 003. [0] L. Ungar, D. Foster, E. Andre, S. Wars, F. S. Wars, D. S. Wars, and J. H. Whspers. Clusterng methods for collaborate flterng. In Workshop on Recommendaton Systems, [1] G.-R. Xue, C. Ln, Q. Yang, W. X, H.-J. Zeng, Y. Y and Z. Chen. Scalable collaborate flterng usng cluster-based smoothng. Proc.the 8th annual nternatonal conference on Research and deelopment n nformaton retreal (SIGIR05), pp ,

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

Adapting Ratings in Memory-Based Collaborative Filtering using Linear Regression

Adapting Ratings in Memory-Based Collaborative Filtering using Linear Regression Adaptng Ratngs n Memory-Based Collaboratve Flterng usng Lnear Regresson Jérôme Kunegs, Şahn Albayrak Technsche Unverstät Berln DAI-Labor Franklnstraße 8/9 10587 Berln, Germany {kunegs,sahn.albayrak}@da-labor.de

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

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

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

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

More information

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

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

A Novel Distributed Collaborative Filtering Algorithm and Its Implementation on P2P Overlay Network*

A Novel Distributed Collaborative Filtering Algorithm and Its Implementation on P2P Overlay Network* A Novel Dstrbuted Collaboratve Flterng Algorthm and Its Implementaton on P2P Overlay Network* Peng Han, Bo Xe, Fan Yang, Jajun Wang, and Rumn Shen Department of Computer Scence and Engneerng, Shangha Jao

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 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

Recommender System based on Higherorder Logic Data Representation

Recommender System based on Higherorder Logic Data Representation Recommender System based on Hgherorder Logc Data Representaton Lnna L Bngru Yang Zhuo Chen School of Informaton Engneerng, Unversty Of Scence and Technology Bejng Abstract Collaboratve Flterng help users

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 Effect of Sparsity on Collaborative Filtering Metrics

The Effect of Sparsity on Collaborative Filtering Metrics The Effect of Sparsty on Collaboratve Flterng Metrcs Jesus Bobadlla & Francsco Serradlla Unversdad Poltecnca de Madrd Crta. De Valenca, Km 7, 8031 Madrd, Span jbob@eu.upm.es, fserra@eu.upm.es Abstract

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

Assessing the Value of Unrated Items in Collaborative Filtering

Assessing the Value of Unrated Items in Collaborative Filtering Assessng the Value of Unrated Items n Collaboratve Flterng Jérôme Kunegs, Andreas Lommatzsch, Martn Mehltz, Şahn Albayrak Technsche Unverstät Berln DAI-Labor Ernst-Reuter-Platz 7, 0587 Berln, Germany {kunegs,andreas.lommatzsch,martn.mehltz,sahn.albayrak}@da-labor.de

More information

Feature-Based Matrix Factorization

Feature-Based Matrix Factorization Feature-Based Matrx Factorzaton arxv:1109.2271v3 [cs.ai] 29 Dec 2011 Tanq Chen, Zhao Zheng, Quxa Lu, Wenan Zhang, Yong Yu {tqchen,zhengzhao,luquxa,wnzhang,yyu}@apex.stu.edu.cn Apex Data & Knowledge Management

More information

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

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

More information

Evolutionary Support Vector Regression based on Multi-Scale Radial Basis Function Kernel

Evolutionary Support Vector Regression based on Multi-Scale Radial Basis Function Kernel Eolutonary Support Vector Regresson based on Mult-Scale Radal Bass Functon Kernel Tanasanee Phenthrakul and Boonserm Kjsrkul Abstract Kernel functons are used n support ector regresson (SVR) to compute

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

EXTENDED BIC CRITERION FOR MODEL SELECTION

EXTENDED BIC CRITERION FOR MODEL SELECTION IDIAP RESEARCH REPORT EXTEDED BIC CRITERIO FOR ODEL SELECTIO Itshak Lapdot Andrew orrs IDIAP-RR-0-4 Dalle olle Insttute for Perceptual Artfcal Intellgence P.O.Box 59 artgny Valas Swtzerland phone +4 7

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

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

CS 534: Computer Vision Model Fitting

CS 534: Computer Vision Model Fitting CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust

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

User Tweets based Genre Prediction and Movie Recommendation using LSI and SVD

User Tweets based Genre Prediction and Movie Recommendation using LSI and SVD User Tweets based Genre Predcton and Move Recommendaton usng LSI and SVD Saksh Bansal, Chetna Gupta Department of CSE/IT Jaypee Insttute of Informaton Technology,sec-62 Noda, Inda sakshbansal76@gmal.com,

More information

Experiments in Text Categorization Using Term Selection by Distance to Transition Point

Experiments in Text Categorization Using Term Selection by Distance to Transition Point Experments n Text Categorzaton Usng Term Selecton by Dstance to Transton Pont Edgar Moyotl-Hernández, Héctor Jménez-Salazar Facultad de Cencas de la Computacón, B. Unversdad Autónoma de Puebla, 14 Sur

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

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

Circuit Analysis I (ENGR 2405) Chapter 3 Method of Analysis Nodal(KCL) and Mesh(KVL)

Circuit Analysis I (ENGR 2405) Chapter 3 Method of Analysis Nodal(KCL) and Mesh(KVL) Crcut Analyss I (ENG 405) Chapter Method of Analyss Nodal(KCL) and Mesh(KVL) Nodal Analyss If nstead of focusng on the oltages of the crcut elements, one looks at the oltages at the nodes of the crcut,

More information

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide

Lobachevsky State University of Nizhni Novgorod. Polyhedron. Quick Start Guide Lobachevsky State Unversty of Nzhn Novgorod Polyhedron Quck Start Gude Nzhn Novgorod 2016 Contents Specfcaton of Polyhedron software... 3 Theoretcal background... 4 1. Interface of Polyhedron... 6 1.1.

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

An Approach for Recommender System by Combining Collaborative Filtering with User Demographics and Items Genres

An Approach for Recommender System by Combining Collaborative Filtering with User Demographics and Items Genres Volume 18 No.13, October 015 An Approach for Recommender System by Combnng Collaboratve Flterng wth User Demographcs and Items Genres Saurabh Kumar Twar Department of Informaton Technology Samrat Ashok

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

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

An Improved Image Segmentation Algorithm Based on the Otsu Method

An Improved Image Segmentation Algorithm Based on the Otsu Method 3th ACIS Internatonal Conference on Software Engneerng, Artfcal Intellgence, Networkng arallel/dstrbuted Computng An Improved Image Segmentaton Algorthm Based on the Otsu Method Mengxng Huang, enjao Yu,

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

Your Neighbors Affect Your Ratings: On Geographical Neighborhood Influence to Rating Prediction

Your Neighbors Affect Your Ratings: On Geographical Neighborhood Influence to Rating Prediction Your Neghbors Affect Your Ratngs: On Geographcal Neghborhood Influence to Ratng Predcton Longke Hu lhu003@e.ntu.edu.sg Axn Sun axsun@ntu.edu.sg Yong Lu luy0054@e.ntu.edu.sg School of Computer Engneerng,

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

Using Fuzzy Logic to Enhance the Large Size Remote Sensing Images

Using Fuzzy Logic to Enhance the Large Size Remote Sensing Images Internatonal Journal of Informaton and Electroncs Engneerng Vol. 5 No. 6 November 015 Usng Fuzzy Logc to Enhance the Large Sze Remote Sensng Images Trung Nguyen Tu Huy Ngo Hoang and Thoa Vu Van Abstract

More information

A Clustering Algorithm Solution to the Collaborative Filtering

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

More information

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

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

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

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

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

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

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices Steps for Computng the Dssmlarty, Entropy, Herfndahl-Hrschman and Accessblty (Gravty wth Competton) Indces I. Dssmlarty Index Measurement: The followng formula can be used to measure the evenness between

More information

A fast algorithm for color image segmentation

A fast algorithm for color image segmentation Unersty of Wollongong Research Onlne Faculty of Informatcs - Papers (Arche) Faculty of Engneerng and Informaton Scences 006 A fast algorthm for color mage segmentaton L. Dong Unersty of Wollongong, lju@uow.edu.au

More information

Description of NTU Approach to NTCIR3 Multilingual Information Retrieval

Description of NTU Approach to NTCIR3 Multilingual Information Retrieval Proceedngs of the Thrd NTCIR Workshop Descrpton of NTU Approach to NTCIR3 Multlngual Informaton Retreval Wen-Cheng Ln and Hsn-Hs Chen Department of Computer Scence and Informaton Engneerng Natonal Tawan

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

FAHP and Modified GRA Based Network Selection in Heterogeneous Wireless Networks

FAHP and Modified GRA Based Network Selection in Heterogeneous Wireless Networks 2017 2nd Internatonal Semnar on Appled Physcs, Optoelectroncs and Photoncs (APOP 2017) ISBN: 978-1-60595-522-3 FAHP and Modfed GRA Based Network Selecton n Heterogeneous Wreless Networks Xaohan DU, Zhqng

More information

Under-Sampling Approaches for Improving Prediction of the Minority Class in an Imbalanced Dataset

Under-Sampling Approaches for Improving Prediction of the Minority Class in an Imbalanced Dataset Under-Samplng Approaches for Improvng Predcton of the Mnorty Class n an Imbalanced Dataset Show-Jane Yen and Yue-Sh Lee Department of Computer Scence and Informaton Engneerng, Mng Chuan Unversty 5 The-Mng

More information

SHAPE RECOGNITION METHOD BASED ON THE k-nearest NEIGHBOR RULE

SHAPE RECOGNITION METHOD BASED ON THE k-nearest NEIGHBOR RULE SHAPE RECOGNITION METHOD BASED ON THE k-nearest NEIGHBOR RULE Dorna Purcaru Faculty of Automaton, Computers and Electroncs Unersty of Craoa 13 Al. I. Cuza Street, Craoa RO-1100 ROMANIA E-mal: dpurcaru@electroncs.uc.ro

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

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

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

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

Abstract. 1 Introduction

Abstract. 1 Introduction Challenges and Solutons for Synthess of Knowledge Regardng Collaboratve Flterng Algorthms Danel Lowd, Olver Godde, Matthew McLaughln, Shuzhen Nong, Yun Wang, and Jonathan L. Herlocker. School of Electrcal

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

Backpropagation: In Search of Performance Parameters

Backpropagation: In Search of Performance Parameters Bacpropagaton: In Search of Performance Parameters ANIL KUMAR ENUMULAPALLY, LINGGUO BU, and KHOSROW KAIKHAH, Ph.D. Computer Scence Department Texas State Unversty-San Marcos San Marcos, TX-78666 USA ae049@txstate.edu,

More information

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

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

More information

A Low-Overhead Routing Protocol for Ad Hoc Networks with selfish nodes

A Low-Overhead Routing Protocol for Ad Hoc Networks with selfish nodes A Low-Oerhead Routng Protocol for Ad Hoc Networks wth selfsh nodes Dongbn Wang 1, Xaofeng Wang 2, Xangzhan Yu 3, Kacheng Q 1, Zhbn Xa 1 1 School of Software Engneerng, Bejng Unersty of Posts and Telecommuncatons,100876,

More information

Keyword-based Document Clustering

Keyword-based Document Clustering Keyword-based ocument lusterng Seung-Shk Kang School of omputer Scence Kookmn Unversty & AIrc hungnung-dong Songbuk-gu Seoul 36-72 Korea sskang@kookmn.ac.kr Abstract ocument clusterng s an aggregaton of

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

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

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

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

Fuzzy Filtering Algorithms for Image Processing: Performance Evaluation of Various Approaches

Fuzzy Filtering Algorithms for Image Processing: Performance Evaluation of Various Approaches Proceedngs of the Internatonal Conference on Cognton and Recognton Fuzzy Flterng Algorthms for Image Processng: Performance Evaluaton of Varous Approaches Rajoo Pandey and Umesh Ghanekar Department of

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

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

An Internal Clustering Validation Index for Boolean Data

An Internal Clustering Validation Index for Boolean Data BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 16, No 6 Specal ssue wth selecton of extended papers from 6th Internatonal Conference on Logstc, Informatcs and Servce Scence

More information

IN recent years, recommender systems, which help users discover

IN recent years, recommender systems, which help users discover Heterogeneous Informaton Network Embeddng for Recommendaton Chuan Sh, Member, IEEE, Bnbn Hu, Wayne Xn Zhao Member, IEEE and Phlp S. Yu, Fellow, IEEE 1 arxv:1711.10730v1 [cs.si] 29 Nov 2017 Abstract Due

More information

[Type text] [Type text] [Type text] Xinquan Yang 1,2 1 School of Computer and Information Engineering, Heze University, Heze, ,

[Type text] [Type text] [Type text] Xinquan Yang 1,2 1 School of Computer and Information Engineering, Heze University, Heze, , [Type text] [Type text] [Type text] ISSN : 0974-7435 Volume 10 Issue 22 BoTechnology 2014 An Indan Journal FULL PAPER BTAIJ, 10(22), 2014 [13798-13803] Research on multdmensonal ndexng based on cloud computng

More information

A Resources Virtualization Approach Supporting Uniform Access to Heterogeneous Grid Resources 1

A Resources Virtualization Approach Supporting Uniform Access to Heterogeneous Grid Resources 1 A Resources Vrtualzaton Approach Supportng Unform Access to Heterogeneous Grd Resources 1 Cunhao Fang 1, Yaoxue Zhang 2, Song Cao 3 1 Tsnghua Natonal Labatory of Inforamaton Scence and Technology 2 Department

More information

Performance Evaluation of Information Retrieval Systems

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

More information

Mining Web Logs with PLSA Based Prediction Model to Improve Web Caching Performance

Mining Web Logs with PLSA Based Prediction Model to Improve Web Caching Performance JOURAL OF COMPUTERS, VOL. 8, O. 5, MAY 2013 1351 Mnng Web Logs wth PLSA Based Predcton Model to Improve Web Cachng Performance Chub Huang Department of Automaton, USTC Key laboratory of network communcaton

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

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

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

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

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

Available online at ScienceDirect. Procedia Environmental Sciences 26 (2015 )

Available online at   ScienceDirect. Procedia Environmental Sciences 26 (2015 ) Avalable onlne at www.scencedrect.com ScenceDrect Proceda Envronmental Scences 26 (2015 ) 109 114 Spatal Statstcs 2015: Emergng Patterns Calbratng a Geographcally Weghted Regresson Model wth Parameter-Specfc

More information

Load-Balanced Anycast Routing

Load-Balanced Anycast Routing Load-Balanced Anycast Routng Chng-Yu Ln, Jung-Hua Lo, and Sy-Yen Kuo Department of Electrcal Engneerng atonal Tawan Unversty, Tape, Tawan sykuo@cc.ee.ntu.edu.tw Abstract For fault-tolerance and load-balance

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

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

Intrinsic Plagiarism Detection Using Character n-gram Profiles

Intrinsic Plagiarism Detection Using Character n-gram Profiles Intrnsc Plagarsm Detecton Usng Character n-gram Profles Efstathos Stamatatos Unversty of the Aegean 83200 - Karlovass, Samos, Greece stamatatos@aegean.gr Abstract: The task of ntrnsc plagarsm detecton

More information

Collaborative Filtering Ensemble for Ranking

Collaborative Filtering Ensemble for Ranking Collaboratve Flterng Ensemble for Rankng ABSTRACT Mchael Jahrer commo research & consultng 8580 Köflach, Austra mchael.ahrer@commo.at Ths paper provdes the soluton of the team commo on the Track2 dataset

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

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

Adaptive Transfer Learning

Adaptive Transfer Learning Adaptve Transfer Learnng Bn Cao, Snno Jaln Pan, Yu Zhang, Dt-Yan Yeung, Qang Yang Hong Kong Unversty of Scence and Technology Clear Water Bay, Kowloon, Hong Kong {caobn,snnopan,zhangyu,dyyeung,qyang}@cse.ust.hk

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

The Effect of Similarity Measures on The Quality of Query Clusters

The Effect of Similarity Measures on The Quality of Query Clusters The effect of smlarty measures on the qualty of query clusters. Fu. L., Goh, D.H., Foo, S., & Na, J.C. (2004). Journal of Informaton Scence, 30(5) 396-407 The Effect of Smlarty Measures on The Qualty of

More information

Scheduling Remote Access to Scientific Instruments in Cyberinfrastructure for Education and Research

Scheduling Remote Access to Scientific Instruments in Cyberinfrastructure for Education and Research Schedulng Remote Access to Scentfc Instruments n Cybernfrastructure for Educaton and Research Je Yn 1, Junwe Cao 2,3,*, Yuexuan Wang 4, Lanchen Lu 1,3 and Cheng Wu 1,3 1 Natonal CIMS Engneerng and Research

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

Study of Data Stream Clustering Based on Bio-inspired Model

Study of Data Stream Clustering Based on Bio-inspired Model , pp.412-418 http://dx.do.org/10.14257/astl.2014.53.86 Study of Data Stream lusterng Based on Bo-nspred Model Yngme L, Mn L, Jngbo Shao, Gaoyang Wang ollege of omputer Scence and Informaton Engneerng,

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

Signed Distance-based Deep Memory Recommender

Signed Distance-based Deep Memory Recommender Sgned Dstance-based Deep Memory Recommender ABSTRACT Personalzed recommendaton algorthms learn a user s preference for an tem, by measurng a dstance/smlarty between them. However, some of exstng recommendaton

More information

Deep Classification in Large-scale Text Hierarchies

Deep Classification in Large-scale Text Hierarchies Deep Classfcaton n Large-scale Text Herarches Gu-Rong Xue Dkan Xng Qang Yang 2 Yong Yu Dept. of Computer Scence and Engneerng Shangha Jao-Tong Unversty {grxue, dkxng, yyu}@apex.sjtu.edu.cn 2 Hong Kong

More information

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics Introducton G10 NAG Fortran Lbrary Chapter Introducton G10 Smoothng n Statstcs Contents 1 Scope of the Chapter... 2 2 Background to the Problems... 2 2.1 Smoothng Methods... 2 2.2 Smoothng Splnes and Regresson

More information

A new selection strategy for selective cluster ensemble based on Diversity and Independency

A new selection strategy for selective cluster ensemble based on Diversity and Independency A new selecton strategy for selectve cluster ensemble based on Dversty and Independency Muhammad Yousefnezhad a, Al Rehanan b, Daoqang Zhang a and Behrouz Mnae-Bdgol c a Department of Computer Scence,

More information