Domain-Constrained Semi-Supervised Mining of Tracking Models in Sensor Networks

Size: px
Start display at page:

Download "Domain-Constrained Semi-Supervised Mining of Tracking Models in Sensor Networks"

Transcription

1 Doman-Constraned Sem-Supervsed Mnng of Trackng Models n Sensor Networks Rong Pan 1, Junhu Zhao 2, Vncent Wenchen Zheng 1, Jeffrey Junfeng Pan 1, Dou Shen 1, Snno Jaln Pan 1 and Qang Yang 1 1 Hong Kong Unversty of Scence & Technology {panrong,vncentz,panjf,dshen,snnopan, qyang}@cse.ust.hk 2 NEC Labs, Chna, Zhongguancun East Road, Bejng 10004, Chna zhaojunhu@research.nec.com.cn ABSTRACT Accurate localzaton of moble objects s a major research problem n sensor networks and an mportant data mnng applcaton. Specfcally, the localzaton problem s to determne the locaton of a clent devce accurately gven the rado sgnal strength values receved at the clent devce from multple beacon sensors or access ponts. Conventonal data mnng and machne learnng methods can be appled to solve ths problem. However, all of them requre large amounts of labeled tranng data, whch can be qute expensve. In ths paper, we propose a probablstc sem-supervsed learnng approach to reduce the calbraton effort and ncrease the trackng accuracy. Our method s based on sem-supervsed condtonal random felds whch can enhance the learned model from a small set of tranng data wth abundant unlabeled data effectvely. To make our method more effcent, we explot a Generalzed EM algorthm coupled wth doman constrants. We valdate our method through extensve experments n a real sensor network usng Crossbow MICA2 sensors. The results demonstrate the advantages of methods compared to other state-of-the-art objecttrackng algorthms. Categores and Subject Descrptors I.2.6 [Artfcal Intellgence]: Learnng; H.2. [Database Management]: Database Applcatons Data mnng General Terms Algorthms Keywords Localzaton, Calbraton, Trackng, Sensor Networks, EM, CRF 1. INTRODUCTION Recently, wreless sensor networks have attracted great nterests n several related research felds and ndustres. Many tasks such as context-aware computng [4] and envronmental montorng can be realzed wth the help of wreless sensor networks, whch offer Permsson to make dgtal or hard copes of all or part of ths work for personal or classroom use s granted wthout fee provded that copes are not made or dstrbuted for proft or commercal advantage and that copes bear ths notce and the full ctaton on the frst page. To copy otherwse, to republsh, to post on servers or to redstrbute to lsts, requres pror specfc permsson and/or a fee. KDD 07, August 12 15, 2007, San Jose, Calforna, USA. Copyrght 2007 ACM /07/000...$5.00. unque advantages of beng lghtweght, dstrbuted, envronmentaware and network-based. Object trackng, event-detecton and actvty recognton can now be realzed n sensor networks usng probablstc algorthms[7, 11]. It s a fundamental task for many of these applcatons to locate moble clent devces usng collected wreless sgnals (n terms of rado-sgnal-strength, RSS, values) from dfferent sensor nodes that act as beacons. In the past, some conventonal data mnng technologes have been appled for solvng the localzaton problem [7, 9]. Generally, some statstcal models are obtaned offlne whch can map sgnals to locatons. These models are then used onlne to predct the clent locatons based on the real-tme sgnal values. Among the past works, many researchers have developed rangng-based algorthms for localzng moble nodes n a wreless sensor network. One approach s multlateraton (e.g., [9]), consstng of two man steps. It frst transforms the sensor readngs nto a dstance measure. It then attempts to recover the coordnate locatons n terms of relatve dstance to the beacon nodes. Ths approach reles on an deal sgnal propagaton model and extensve hardware support. It suffers from low accuracy problem snce RSSs do not follow deal propagaton patterns, especally n complex envronments. In ths paper, we address ths problem usng a sem-supervsed statstcal relatonal learnng approach based on condtonal random felds (Sem-CRF). We assume that a moble sensor node moves n a sensor network envronment. The RSS of the moble node can be receved by several sensors n the network, whch are then forwarded to a processor for trackng. It can also happens n the way that all sensors n the network are sendng sgnals to the moble sensor node, whch performs the localzaton tself. In ether case, we have a sequence of mult-dmensonal vectors that corresponds to a trace. Each vector along the trace can be labeled wth a physcal locaton coordnate, or unlabelled. Ths paper makes the followng contrbutons. Frst, we dentfy and solve a major bottleneck n the applcaton of data mnng technologes n sensor networks. Second, we present a novel semsupervsed learnng method for moble-node trackng and localzaton by utlzng both labeled and unlabelled RSS trace data. Thrd, we ntroduce doman-drven heurstcs for reducng the complexty of the learnng procedure, whch greatly mprove the scalablty of the statstcal models. Fnally, we valdate the proposed methods through the experments over a real sensor network. 2. THE Sem-CRF ALGORITHM FOR TRACKING AND LOCALIZATION 2.1 Problem Statement Consder a two-dmensonal trackng problem. Our objectve s 1023

2 to determne the locaton of a moble object y t =(u t,v t) C as t moves n a sequence, gven the observed sgnal vectors x t. Fgure 3 shows an example of the floor n one of our expermental test beds, whch conssts of N =beacon nodes and one moble unknown node. The localzaton problem can be converted to a supervsed classfcaton problem f we had suffcent labeled data for each locaton. However, when the labeled data are nsuffcent at each locaton, we wsh to make the best use of some partally labelled or totally unlabeled RSS sequences as well n our predcton. Now let us formally ntroduce the notaton of tranng data. In our study, the tranng data consst of a set of fully labelled sequences D f = {(X 1,Y 1),...,(X M,Y M )}, and a set of partally labeled (or totally unlabeled) sequences D p = {(X M+1,Y M+1),...,(X M+L,Y M+L)},whereX s a sequence of sgnal vectors x 1,...,x m, =1,...,M + L,andY s a sequence of correspondng locatons y 1,...,y m, =1,...,M + L. Some values of y j are unknown for M +1 M + L. A moble robot can be employed to collect these unlabelled data by smply wanderng around. 2.2 Lnear-chan CRF The CRF Model In ths paper, we propose a statstcal relatonal learnng approach usng CRF to explot the relatonshp between RSS readngs at two neghborng tme ponts n terms of ther correspondng physcal dstance. As a moble object moves around, a sequence of RSS values can be receved, wth each correspondng to a certan locaton. Ths process can be modeled by an 1-D Lnear-chan CRF as ntroduced n the followng secton, where the states correspond to the locaton labels and the nputs or observatons correspond to the RSS readngs. Lnear-chan CRF models have been wdely used to model sequental data. These models can be roughly vewed as condtonally-traned fnte state machnes [5]. A lnear-chan CRF, as shown n Fgure 1, defnes a dstrbuton over state sequence y = y 1,y 2,...,y T gven an nput sequence x = x 1,x 2,...,x T by makng a frst-order Markov assumpton on states, where T s Unobserved Y Observed X y 1 y 2 x 1 x 2 Fgure 1: Dagram of Lnear-chan CRF the length of the sequence. These Markov assumptons mply that the dstrbuton over sequences factorzes n terms of potental functons Φ t(y t 1,y t, x) as: p(y x) = Q t y T x T Φt(yt 1,yt, x), (1) Z(x) where the partton functon Z(x) s the normalzaton constant that makes the probablty of all state sequences sum to one. It s defned as follows: Z(x) = X Y Φ t(y t 1,y t, x). (2) y t The potental functons Φ t(y t 1,y t, x) can be nterpreted as the cost of makng a transton from state y t 1 to state y t at tme t, smlar to a transton probablty n an HMM. Computng the partton functon Z(x) requres summng over the exponentally many possble state sequences y. By explotng the Markov assumpton, however, Z(x) (as well as the node margnal p(y t x) and the Vterb labelng) can be calculated effcently by varants of the standard dynamc programmng algorthms used n HMM. We assume that potentals factorze themselves accordng to a set of features f k that are gven and fxed, so that! Φ(y t 1,y t, x) =exp λ k f k (y t 1,y t, x,t). (3) X k The model parameters are a set of real weghts Λ = {λ k }, one for each feature (to be defned below). The feature functons can descrbe any aspect of a transton from y t 1 to y t as well as y t and the global characterstcs of x. For example, f k may have value 1 when the dstance between y t 1 and y t s smaller than 50cm Parameter Estmaton The parameters Λ can be estmated through a maxmum lkelhood procedure usng the tranng data. That s, we can estmate them by maxmzng the condtonal log-lkelhood of the labeled sequences n the tranng data Ψ={(X 1,Y 1),...,(X M,Y M ))}, whch s defned as: MX L(Λ) = log(p (Y X ;Λ)), (4) =1 where M s the number of sequences. As dscussed n Sutton et al. n [10], L(Λ) s concave n lght of the convexty of the knd of functons g(x) =log P exp x Inference Gven the condtonal probablty of the state sequence defned by a CRF n Equaton (1) and the parameters Λ, the most probable labelng sequence can be obtaned as Y =argmaxp (Y X;Λ), (5) Y whch can be effcently calculated usng the Vterb algorthm []. The margnal probablty of states at each poston n the sequence can be computed by a dynamc programmng nference procedure smlar to the forward-backward procedure for HMM [3]. We can defne the forward values α t(y x) by settng α 1(y x) equal to the probablty of startng wth state y and then terate as follows: α t(y x) = X y α t 1(y x)exp `Λ t(y,y,x), (6) where Λ t(y,y,x) s defned by: Λ t(y,y,x) = X k Then Z(x) equals to P y αt(y x). β (y x) can be defned smlarly. λ k f k (y t 1 = y,y t = y, x) (7) The backward values β t(y x) = X y β t+1(y x)exp `Λ t(y, y, x). () After that, we calculate the margnal probablty of each locaton gven the observed sgnal sequence: P (y t = g x) = αt(g x) βt(g x). (9) Z(x) So far, we have ntroduced a lnear-chan CRF model for unknown moble-node trackng. We can see that f k (y t 1,y t, x) (n 1024

3 Equaton (3)) s an arbtrary feature functon over the entre observed sequences and the states at postons t and t 1. In our problem, the locatons are two-dmensonal contnuous values. The number of possble locatons are nfnte large. Therefore, t s extremely dffcult to compute the feature of two arbtrary locatons. Fortunately, the trackng area s known n advance usually. One soluton s to dscretze a 2-D locaton space nto grds. For nstance, n a 5m 4m area, we can dvde t nto 10 grds wth each grd beng 50 50cm 2. Ths example s shown n Fgure 2. In ths way, we can convert the known locatons nto such grds. In the test phase, f a moble object s located at grd g, we can use the coordnates of the center pont n g to represent the locaton of the moble object. After lmtng the locaton space, t s possble to use the lnear-chan CRF approach for trackng problem. However, a major ssue s how to determne the sze of the grd. Ths problem can be solved n two ways. Frst, the sze often s determned by the nature of the problem tself, whch s decded by the precson requrement posed by applcaton users. Another approach s to study the problem emprcally, as we wll do n the expermental secton. g 1 g 2 g 24 g 35 g 46 g g 9 Fgure 2: A demo of reducton of locatons to grds Incorporatng Doman Constrants After reducng locatons to grds, we can specalze the feature functons for each possble transton among dfferent grds. That s, we can defne f k (y t 1 = g, y t = h, x) by f (g,h) (t 1,t,x). However, the number of the transton feature functons, as well as the correspondng parameters, reaches n 2 (n s the number of grds), whch can be qute large. For nstance, n the above example n Fgure 2, n = 0, then n the CRF learnng, we need to estmate 6400 parameters for the potental f k (y t 1,y t, x). Although we can stll estmate the values of the parameters wth large n, t wll certanly ncrease the computatonal cost and run the rsk of overfttng. What s worse, learnng CRF wth more parameters requres more tranng data, whch wll ncrease the labellng costs. In addton, we also need to trade off the complexty of the model and ts generalzaton capablty. If we ncrease the grd sze to reduce the computatonal cost, we wll sacrfce the estmaton accuracy. In ths paper, we ncorporate the doman constrants n the data mnng process to reduce the number of parameters that need be learned. In partcular, we note that a moble object n a sensor network typcally moves around n the same way, such that the lkelhood of transtng between two neghborng ponts are roughly same. The lkelhood of travelng between two dstant ponts wll also be roughly the same, although the value wll be much smaller. Such a doman constrant s supported by our experments. g 0 To ncorporate the doman constrants mentoned above, we use a so-called parameter tyng technque that s desgned for speech recognton [6] to combne smlar parameters. Our assumpton s that the characterstcs of two transtons wth the same dstance are alke. Intutvely, n Fgure 2, we observe that the transtons g 1 g 2 and g g 9 should happen wth smlar frequences as they both transt by one grd n terms of Eucldean dstance. Smlarly, the transtons g 24 g 35 and g 35 g 46 should happen smlarly as the ther Eucldean dstances are both 2 grds. From ths observaton, we can te the parameters of the transton feature functons so long as they have the same transton dstance, whch s defned as follows: The value of the transton feature functon f k (y t 1,y t, x) equals one f and only f ds(y t 1,y t)=k, where the ds defnes the dstance between the two ponts. As expected, the number of parameters s greatly reduced by usng ths constrant. 2.3 The Sem-CRF Algorthm In ths secton, we ntroduce how to ncorporate sequences whose labels are fully or partally observed n the parameter estmaton of CRF. An effcent method for parameter estmaton wth ncomplete data can be derved by the extenson of EM algorthm [2]. In ths paper, we use a Generalzed Expectaton Maxmzaton (GEM) algorthm to learn the parameters Λ of CRF wth both fully and partally observed data [1]. In the GEM algorthm, the probablstc optmzaton problem s dvded nto two-step teratons. The unobserved data are estmated n the E-step wth the parameters obtaned n the last teraton and the parameters of CRF are optmzed n the M-step. We frst compute the log lkelhood of Equaton (4) wth expectaton over the unobserved data as follows: L(Λ; Λ t ) = P M =1 log P (Y X;Λ)+ P M+L P P (Y (u) X,Y (o) ;Λ t )logp(y (u) = P M =M+1 Y (u) P=1 M+L =M+1,Y (o) Pt Λt(y,y,X ) P M+L =1 log Z(X P )+ P (Y (u) X,Y (o) ;Λ t ) P t Λt(y,y,X ) Y (u) In ths equaton, Y (u) s the unobserved locatons of the -th sequence, Y (o) s the observed counterpart, and Λ t(y,y,x )= X k λ kf k (y,y,x ), Z(X )= X y exp Xt Λt(yt 1,yt,X). Smlar to Equaton (4), L(Λ; Λ t ) s also concave. We can use the same method to optmze t. The only problem left s how to nfer for partally observed sequences. We need to change Equatons (6) and () for some cases. If y t = j s observed, we drectly assgn 1 to α t(y = j x) and β t(y = j x) and assgn 0 to the other values of α t and β t.ify t s not observed, we follow Equatons (6) and (). The new nference formulae are summarzed n Equatons (10) to (12). >< 0 y t j α t(y = j x) = 1 y t = j >: P y α t 1(y x)exp (Λ t(y,y,x)) y t unseen (10) X ;Λ) 1025

4 >< 0 y t j β t(y = j x) = 1 y t = j >: P y β t+1(y x)exp (Λ t(y,y,x)) y t unseen (11) P (y t = k x) = αt(y = k x) βt(y = k x) Z x. (12) We now summarze the Sem-CRF learnng algorthm n Table 1. There are several ways of parameter ntalzaton. The common one s to randomly assgn them values from 0 to 1. To speed up the convergence, we use an alternatve that prelmnarly estmates parameters wth labeled data. As to the number of teratons, we wll dscuss t n the expermental secton. Table 1: The tranng algorthm for CRF wth both fully and partally observed data. Algorthm Sem-CRF Input: The fully and partally observed data D f, D p Output: The parameters Λ of CRF 1 Intalze parameters Λ 0 of CRF. Λ t =Λ 0. whle log-lkelhood has not converged or the max number of teratons s not reached, do % ====== E-step ====== Compute the expectatons of the all unobserved locatons, by Equatons (10) to (12). % ====== M-step ====== Optmze Λ usng L-BFGS. Λ t =Λ. endwhle return Λ Camera-1 2 Camera Moble Node 7 Projected Plane Real Plane Camera-3 5 Camera-2 Beacon Node Fgure 3: Expermental Test-Bed 4 MICA2 3. EXPERIMENTAL EVALUATION MICA2Dot Sony AIBO LEGO Mndstorms 3.1 Expermental Setup We test the effectveness and robustness of our locaton trackng algorthm for moble sensor nodes n a sensor network based on the RSS sgnals. Our experments are performed n the Pervasve Computng Laboratory (Fgure 3) n the Department of Computer Scence and Engneerng at Hong Kong Unversty of Scence and Technology. The room s large enough for us to set up an expermental test-bed of 5.0 meters by 4.0 meters. In Fgure 3, P 1P 3 = P 4P 6 =5.0m and P 1P 4 = P 3P 6 =4.0m. There are three man components of our setup: Wreless Sensor Networks. We use CrossBow MICA2 and MICA2Dot to construct a wreless sensor network. We program these sensor nodes to broadcast and detect beacon frames perodcally so that they could measure the RSS of each other. Moble Robots. We try dfferent knds of robots that can run freely around the floor at dfferent speeds, such as a Sony AIBO dogs, LEGO Mndstorms and off-the-shelf toy cars. Fgure 3 shows that a sensor node s attached on top of a toy car whch can be remotely controlled by rado at the speed of 0.4 m/s. A Camera Array s used to record the ground truth locatons of the moble robots for our tranng and test data. We use two performance measurements to evaluate the orgnal CRF and the CRF model usng parameter tyng (denoted by CRF- PT) localzaton algorthms. The frst metrc s the mean errordstance values between estmated and true locatons. The second measurement s the accuracy n percentage. Gven an error-dstance threshold θ, the accuracy rate s the probablty that the dstance between the estmated and true locatons s smaller than θ. Two more baselnes n our experments nclude (1) Logstc Regresson (LR), (2) Support Vector Regresson (SVR). We control a moble robot to run and stop around the test area (Fgure 3) for collectng data wth samplng nterval 0.5s. The data set formed a trace of length about 600m wth 3, 000 examples. For every experment below, we randomly select a subset of the data as fully observed tranng data, a subset of data as partally observed tranng data by randomly removng the locatons assocated wth them, and evaluate the performance on the rest. To reduce the statstcal varablty, we repeated the experments for 30 tmes and reported the average results. 3.2 Convergence of Sem-CRF One queston about the the Sem-CRF algorthm s the convergence of the EM teratons. In ths experment, we use 10 fully labelled and 50 partally labelled sequental data to tran the CRF, where the length of each sequence s 5 and only one node s labelled n the partally labelled data. Fgure 4 shows the convergence rate of Sem-CRF. We can see that about 4 teratons are enough. In the experments of ths paper, the maxmum number of teratons of the Sem-CRF s set to 10. Summaton of Log Lkelhooh Number of Iteratons Fgure 4: Convergence rate of Sem-CRF 1026

5 3.3 Sem-CRF vs. Baselnes In the followng experments, we fx the tranng data sze at 550, and tune the rato of the labelled data from 0.33 to 1. In Fgure 5, we show the mean error performance of the four algorthms descrbed above. As can be seen, Sem-CRF consstently outperforms the other algorthms n terms of mean error dstance, whle CRF beats the remanng two baselnes. One mportant reason s they both effectvely leverage the sequental nformaton of the moble node. Moreover, as Sem-CRF can also learn from the unlabelled data, t gans much better performance when there are a lot of such data wth a small porton of labelled ones. We lst some more nformaton of these experments ncludng the accuracy performance n Table 2. Mean Error (cm) LR SVR CRF GEM CRF proporton of labelled nstances Fgure 5: Vary the rato of tranng set sze Table 2: Performance of the tested approaches Approach Mean(cm) Accuracy at 100cm Sem-CRF % CRF % SVR % LR % 3.4 Impact of Grd Szes The grd sze may affect the performance of the localzaton algorthms. In ths experment, we fx the rato of labelled data at 5% and vary the sde length of the grds from 10cm to 100cm. Fgure 6 shows the expermental results of CRF and Sem-CRF. From the fgure we can see that when the grd sze ranges from 20cm to 50cm, the performance of both the two methods s less senstve than that wth the grd sze of 10cm and 100cm. Mean Error (cm) CRF GEM CRF Grd Sze (cm) Fgure 6: Vary the grd sze (rato of labelled data s 5%.) 4. CONCLUSION AND FUTURE WORKS We have presented a new approach to reducng the calbraton effort when trackng moble sensor nodes n a wreless sensor network. Our approach made extensve use of the sequental nformaton of movng sensor s trajectory. These sequences provded unlabelled examples whch can be used to tran CRF together wth the manually labelled RSS values. We ntroduce a Sem-CRF model to utlze such partally labelled data. By usng parameter tyng technques we sgnfcantly mprove the performance of Sem-CRF algorthm whle reducng calbraton effort. A sensor network was set up based on Crossbow MICA2 and MICA2Dot nodes whch are used as both beacon and moble nodes. Expermental results showed that the proposed method could acheve a better performance wth relatvely fewer number of labelled examples. In the future, we plan to contnue to test the Sem-CRF based framework n a large scale sensor network. We are also nterested n ntroducng dfferent factors, such as changng tme and space, to see how the knowledge learned n one settng can be appled to another. 5. REFERENCES [1] H. L. Cheu,, S. W. Lee, and P. K. Lesle. Actvty recognton from physologcal data usng condtonal random felds, January [2] A. P. Dempster, N. M. Lard, and D. Rubn. Maxmum lkelhood from ncomplete data va the em algorthm. Journal of the Royal Statstcal Socety, 1(39):1 3, [3] J. Lafferty, A. McCallum, and F. Perera. Condtonal random felds: Probablstc models for segmentng and labelng sequence data. In Proc. 1th Internatonal Conf. on Machne Learnng, pages Morgan Kaufmann, San Francsco, CA, [4] J. Lester, T. Choudhury, N. Kern, G. Borrello, and B. Hannaford. A hybrd dscrmnatve/generatve approach for modelng human actvtes. In IJCAI, pages , [5] A. McCallum. Effcently nducng features of condtonal random felds. In C. Meek and U. Kjærulff, edtors, UAI, pages Morgan Kaufmann, [6] C. Neukrchen, D. Wllett, and G. Rgoll. Soft State-Tyng for HMM-Based Speech Recognton. In 5th Internatonal Conference on Spoken Language Processsng (ICSLP), pages , Sydney, 199. [7] X. Nguyen, M. I. Jordan, and B. Snopol. A kernel-based learnng approach to ad hoc sensor network localzaton. ACM Transactons on Sensor Networks, 1(1): , [] L. R. Rabner. A tutoral on hdden markov models and selected applcatons n speech recognton. Proceedngs of the IEEE, 77(2):257 26, 199. [9] A. Savvdes, C. Han, and M. B. Strvastava. Dynamc fne-graned localzaton n ad-hoc networks of sensors. In Proceedngs of the 7th Annual Internatonal Conference on Moble Computng and Networkng, pages , Rome, Italy, [10] C. Sutton and A. McCallum. An ntroducton to condtonal random felds for relatonal learnng. In L. Getoor and B. Taskar, edtors, Introducton to Statstcal Relatonal Learnng. MIT Press, [11] J. Yn, X. Cha, and Q. Yang. Hgh-level goal recognton n a wreless LAN. In Proceedngs of the Nneteenth Natonal Conference on Artfcal Intellgence, pages 57 54, San Jose, CA, USA, July

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

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

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

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

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

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

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

More information

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

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

More information

The 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

Classifier Selection Based on Data Complexity Measures *

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

More information

An 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

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

Fast and Scalable Training of Semi-Supervised CRFs with Application to Activity Recognition

Fast and Scalable Training of Semi-Supervised CRFs with Application to Activity Recognition Fast and Scalable Tranng of Sem-Supervsed CRFs wth Applcaton to Actvty Recognton Maryam Mahdavan Computer Scence Department Unversty of Brtsh Columba Vancouver, BC, Canada Tanzeem Choudhury Intel Research

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

Three supervised learning methods on pen digits character recognition dataset

Three supervised learning methods on pen digits character recognition dataset Three supervsed learnng methods on pen dgts character recognton dataset Chrs Flezach Department of Computer Scence and Engneerng Unversty of Calforna, San Dego San Dego, CA 92093 cflezac@cs.ucsd.edu Satoru

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

Positive Semi-definite Programming Localization in Wireless Sensor Networks

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

More information

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

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

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

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

More information

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

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

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

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

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

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

Joint Recognition of Multiple Concurrent Activities using Factorial Conditional Random Fields

Joint Recognition of Multiple Concurrent Activities using Factorial Conditional Random Fields Jont Recognton of Multple Concurrent Actvtes usng Factoral Condtonal Random Felds Tsu-yu Wu and Cha-chun Lan and Jane Yung-jen Hsu Department of Computer Scence and Informaton Engneerng Natonal Tawan 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

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

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

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

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

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

Application of Maximum Entropy Markov Models on the Protein Secondary Structure Predictions

Application of Maximum Entropy Markov Models on the Protein Secondary Structure Predictions Applcaton of Maxmum Entropy Markov Models on the Proten Secondary Structure Predctons Yohan Km Department of Chemstry and Bochemstry Unversty of Calforna, San Dego La Jolla, CA 92093 ykm@ucsd.edu Abstract

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

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

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

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

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

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

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

Classification / Regression Support Vector Machines

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

More information

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

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

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

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

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

More information

Concurrent Apriori Data Mining Algorithms

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

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 256 a course in differential equations for engineering students Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the

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

Overview. Basic Setup [9] Motivation and Tasks. Modularization 2008/2/20 IMPROVED COVERAGE CONTROL USING ONLY LOCAL INFORMATION

Overview. Basic Setup [9] Motivation and Tasks. Modularization 2008/2/20 IMPROVED COVERAGE CONTROL USING ONLY LOCAL INFORMATION Overvew 2 IMPROVED COVERAGE CONTROL USING ONLY LOCAL INFORMATION Introducton Mult- Smulator MASIM Theoretcal Work and Smulaton Results Concluson Jay Wagenpfel, Adran Trachte Motvaton and Tasks Basc Setup

More information

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

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

More information

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

Biostatistics 615/815

Biostatistics 615/815 The E-M Algorthm Bostatstcs 615/815 Lecture 17 Last Lecture: The Smplex Method General method for optmzaton Makes few assumptons about functon Crawls towards mnmum Some recommendatons Multple startng ponts

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

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

The Codesign Challenge

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

More information

Virtual Machine Migration based on Trust Measurement of Computer Node

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

More information

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

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

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

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

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

Cost-efficient deployment of distributed software services

Cost-efficient deployment of distributed software services 1/30 Cost-effcent deployment of dstrbuted software servces csorba@tem.ntnu.no 2/30 Short ntroducton & contents Cost-effcent deployment of dstrbuted software servces Cost functons Bo-nspred decentralzed

More information

Accurate Information Extraction from Research Papers using Conditional Random Fields

Accurate Information Extraction from Research Papers using Conditional Random Fields Accurate Informaton Extracton from Research Papers usng Condtonal Random Felds Fuchun Peng Department of Computer Scence Unversty of Massachusetts Amherst, MA 01003 fuchun@cs.umass.edu Andrew McCallum

More information

Local Quaternary Patterns and Feature Local Quaternary Patterns

Local Quaternary Patterns and Feature Local Quaternary Patterns Local Quaternary Patterns and Feature Local Quaternary Patterns Jayu Gu and Chengjun Lu The Department of Computer Scence, New Jersey Insttute of Technology, Newark, NJ 0102, USA Abstract - Ths paper presents

More information

CS246: Mining Massive Datasets Jure Leskovec, Stanford University

CS246: Mining Massive Datasets Jure Leskovec, Stanford University CS46: Mnng Massve Datasets Jure Leskovec, Stanford Unversty http://cs46.stanford.edu /19/013 Jure Leskovec, Stanford CS46: Mnng Massve Datasets, http://cs46.stanford.edu Perceptron: y = sgn( x Ho to fnd

More information

Detection of an Object by using Principal Component Analysis

Detection of an Object by using Principal Component Analysis Detecton of an Object by usng Prncpal Component Analyss 1. G. Nagaven, 2. Dr. T. Sreenvasulu Reddy 1. M.Tech, Department of EEE, SVUCE, Trupath, Inda. 2. Assoc. Professor, Department of ECE, SVUCE, Trupath,

More information

IMPACT OF RADIO MAP SIMULATION ON POSITIONING IN INDOOR ENVIRONTMENT USING FINGER PRINTING ALGORITHMS

IMPACT OF RADIO MAP SIMULATION ON POSITIONING IN INDOOR ENVIRONTMENT USING FINGER PRINTING ALGORITHMS IMPACT OF RADIO MAP SIMULATION ON POSITIONING IN INDOOR ENVIRONTMENT USING FINGER PRINTING ALGORITHMS Jura Macha and Peter Brda Unversty of Zlna, Faculty of Electrcal Engneerng, Department of Telecommuncatons

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

High-Boost Mesh Filtering for 3-D Shape Enhancement

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

More information

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

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

Anytime Predictive Navigation of an Autonomous Robot

Anytime Predictive Navigation of an Autonomous Robot Anytme Predctve Navgaton of an Autonomous Robot Shu Yun Chung Department of Mechancal Engneerng Natonal Tawan Unversty Tape, Tawan Emal shuyun@robot0.me.ntu.edu.tw Abstract To acheve fully autonomous moble

More information

What s Next for POS Tagging. Statistical NLP Spring Feature Templates. Maxent Taggers. HMM Trellis. Decoding. Lecture 8: Word Classes

What s Next for POS Tagging. Statistical NLP Spring Feature Templates. Maxent Taggers. HMM Trellis. Decoding. Lecture 8: Word Classes Statstcal NLP Sprng 2008 Lecture 8: Word Classes Dan Klen UC Berkeley What s Next for POS Taggng Better features! RB PRP VBD IN RB IN PRP VBD. They left as soon as he arrved. We could fx ths wth a feature

More information

Air Transport Demand. Ta-Hui Yang Associate Professor Department of Logistics Management National Kaohsiung First Univ. of Sci. & Tech.

Air Transport Demand. Ta-Hui Yang Associate Professor Department of Logistics Management National Kaohsiung First Univ. of Sci. & Tech. Ar Transport Demand Ta-Hu Yang Assocate Professor Department of Logstcs Management Natonal Kaohsung Frst Unv. of Sc. & Tech. 1 Ar Transport Demand Demand for ar transport between two ctes or two regons

More information

Fusion Performance Model for Distributed Tracking and Classification

Fusion Performance Model for Distributed Tracking and Classification Fuson Performance Model for Dstrbuted rackng and Classfcaton K.C. Chang and Yng Song Dept. of SEOR, School of I&E George Mason Unversty FAIRFAX, VA kchang@gmu.edu Martn Lggns Verdan Systems Dvson, Inc.

More information

Parallel matrix-vector multiplication

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

More information

Face Detection with Deep Learning

Face Detection with Deep Learning Face Detecton wth Deep Learnng Yu Shen Yus122@ucsd.edu A13227146 Kuan-We Chen kuc010@ucsd.edu A99045121 Yzhou Hao y3hao@ucsd.edu A98017773 Mn Hsuan Wu mhwu@ucsd.edu A92424998 Abstract The project here

More information

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

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

More information

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

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

Learning to Project in Multi-Objective Binary Linear Programming

Learning to Project in Multi-Objective Binary Linear Programming Learnng to Project n Mult-Objectve Bnary Lnear Programmng Alvaro Serra-Altamranda Department of Industral and Management System Engneerng, Unversty of South Florda, Tampa, FL, 33620 USA, amserra@mal.usf.edu,

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

Maintaining temporal validity of real-time data on non-continuously executing resources

Maintaining temporal validity of real-time data on non-continuously executing resources Mantanng temporal valdty of real-tme data on non-contnuously executng resources Tan Ba, Hong Lu and Juan Yang Hunan Insttute of Scence and Technology, College of Computer Scence, 44, Yueyang, Chna Wuhan

More information

Active Contours/Snakes

Active Contours/Snakes Actve Contours/Snakes Erkut Erdem Acknowledgement: The sldes are adapted from the sldes prepared by K. Grauman of Unversty of Texas at Austn Fttng: Edges vs. boundares Edges useful sgnal to ndcate occludng

More information

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

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

More information

Efficient Distributed File System (EDFS)

Efficient Distributed File System (EDFS) Effcent Dstrbuted Fle System (EDFS) (Sem-Centralzed) Debessay(Debsh) Fesehaye, Rahul Malk & Klara Naherstedt Unversty of Illnos-Urbana Champagn Contents Problem Statement, Related Work, EDFS Desgn Rate

More information

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 1. SSDH: Semi-supervised Deep Hashing for Large Scale Image Retrieval

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 1. SSDH: Semi-supervised Deep Hashing for Large Scale Image Retrieval IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY SSDH: Sem-supervsed Deep Hashng for Large Scale Image Retreval Jan Zhang, and Yuxn Peng arxv:607.08477v2 [cs.cv] 8 Jun 207 Abstract Hashng

More information

Wireless Sensor Network Localization Research

Wireless Sensor Network Localization Research Sensors & Transducers 014 by IFSA Publshng, S L http://wwwsensorsportalcom Wreless Sensor Network Localzaton Research Lang Xn School of Informaton Scence and Engneerng, Hunan Internatonal Economcs Unversty,

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

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

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

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 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

3D vector computer graphics

3D vector computer graphics 3D vector computer graphcs Paolo Varagnolo: freelance engneer Padova Aprl 2016 Prvate Practce ----------------------------------- 1. Introducton Vector 3D model representaton n computer graphcs requres

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

Hermite Splines in Lie Groups as Products of Geodesics

Hermite Splines in Lie Groups as Products of Geodesics Hermte Splnes n Le Groups as Products of Geodescs Ethan Eade Updated May 28, 2017 1 Introducton 1.1 Goal Ths document defnes a curve n the Le group G parametrzed by tme and by structural parameters n the

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

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

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

More information

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

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