PSO Algorithm Particle Filters for Improving the Performance of Lane Detection and Tracking Systems in Difficult Roads

Size: px
Start display at page:

Download "PSO Algorithm Particle Filters for Improving the Performance of Lane Detection and Tracking Systems in Difficult Roads"

Transcription

1 Sensors 2012, 12, ; do: /s Arcle OPEN ACCESS sensors ISSN PSO Algorhm Parcle Flers for Improvng he Performance of Lane Deecon and Trackng Sysems n Dffcul Roads Wen-Chang Cheng Deparmen of Compuer Scence and Informaon Engneerng, Chaoyang Unversy of Technology, 168, Jfong E. Rd., Wufong Dsrc, Tachung 41349, Tawan; E-Mal: wccheng@cyu.edu.w; Tel.: (ex. 5208); Fax: Receved: 9 Augus 2012; n revsed form: 7 December 2012 / Acceped: 10 December 2012 / Publshed: 12 December 2012 Absrac: In hs paper we propose a robus lane deecon and rackng mehod by combnng parcle flers wh he parcle swarm opmzaon mehod. Ths mehod manly uses he parcle flers o deec and rack he local opmum of he lane model n he npu mage and hen seeks he global opmal soluon of he lane model by a parcle swarm opmzaon mehod. The parcle fler can effecvely complee lane deecon and rackng n complcaed or varable lane envronmens. However, he resul obaned s usually a local opmal sysem saus raher han he global opmal sysem saus. Thus, he parcle swarm opmzaon mehod s used o furher refne he global opmal sysem saus n all sysem sauses. Snce he parcle swarm opmzaon mehod s a global opmzaon algorhm based on erave compung, can fnd he global opmal lane model by smulang he food fndng way of fsh school or nsecs under he muual cooperaon of all parcles. In verfcaon esng, he es envronmens ncluded hghways and ordnary roads as well as sragh and curved lanes, uphll and downhll lanes, lane changes, ec. Our proposed mehod can complee he lane deecon and rackng more accuraely and effecvely hen exsng opons. Keywords: drver asssance sysem; lane deparure warnng sysem; lane followng sysem; lane collson warnng sysem; lane model

2 Sensors 2012, Inroducon Lane deecon plays an mporan role n many drver asssance sysems such as lane deparure warnng sysems, lane followng sysems, lane collson warnng sysems, ec. [1 6]. A lane deparure warnng sysem manly deermnes wheher he vehcle s devang from he curren lane or crosses he lane based on he geomercal relaonshp beween he lane and he vehcle. When he vehcle depars from or crosses he lane, a warnng s gven for he frs me o draw he drver s aenon or he urn sgnal s auomacally acvaed when changng lanes; hs can effecvely reduce he ncdence of accdens. However, he lane followng sysem s usually used o monor wheher he drver s focused on drvng or no. When he drver s absen-mnded, he vehcle can easly drf o he lef and rgh. Therefore, he drvng saus can be known hrough a lane followng sysem. As for he lane collson warnng sysem, f he sysem does no know he lane locaon, mgh gve wrong warnngs by msakng he barrers on he roadsde as obsacles n he lane. Therefore, he lane deecon can effecvely reduce he rae of false alarms. From he above dscusson can be concluded ha lane deecon ndeed plays an mporan role n drvng asssance sysems. The commonly used mehod n proposed lane deecon mehods for lane rackng s he Kalman fler [7 15]. I s a hgh-effcency recurson fler whch can esmae he saus of dynamc sysems based on a seres of ncomplee measuremens conanng nerference nformaon. The dynamc sysem s expressed as a hdden Markov model and s saus s used o express he lane models and carry ou lane deecon and rackng. One of s advanages s s ably o speed up lane deecon. I can ncrease he lane search speed and accuracy n npu mages snce he sysem esablshes he area of neres wh lane models. Anoher advanage s ha can smooh he deecon resuls, reduce he errors and ncrease he sably of lane deecon. However, all hese advanages become problems n ceran road suaons, for example, dsconnuous lane markngs, lane changes and complcaed lane condons. Snce he range of area of neres lms he deecon, he above suaons can easly cause deecon errors. The soluon s o use he parcle fler (PF) o carry ou lane deecon and rackng [16 21]. The parcle fler s a novel echnology for probably-based rackng. I s a mehod based on he saus of dynamc sysem wh mulple Gaussan probably dsrbuon funcons, whch can sll effecvely complee lane deecon and rackng under condons such as dsconnuous lanes, lane changes and complcaed lanes. A parcle-based lane deecon and rackng approach usually sars wh parcle predcon, measuremen, and re-samplng [1]. Souhall and Taylor [16] presened a lane deecor based on a condensaon framework [22], whch uses markng pons as measuremen feaures. Each pon n he mage receves a score based on he dsance o he neares lane markng, and hese scores are used o compue he machng score of each parcle. The measuremen sep s consderably smpler han he Kalman fler, because usually consss of a comparson beween he parcle and he mage daa. In addon, many dfferen cues are used n he measuremen seps of he parcle fler [17 20] so as o ncrease he accuracy of he measuremen seps, for example, edge pons, color, emplae, sereo vson, ec. Danescu and Nedevsch [21] used sereo vson o help confrm he lane markng pons and used he condensaon parcle fler o acheve lane deecon and rackng. Even so, he parcle fler generally akes he parcle wh he bgges wegh or weghed sum of all parcles as he las oupu parcle. Therefore, he oupu resul s a local opmal soluon raher han a global opmal soluon n he problem doman. Ths paper proposes a new PSO-PF

3 Sensors 2012, algorhm by combnng he parcle fler and parcle swarm opmzaon (PSO) mehod. Such an algorhm s an erave operaon opmzaon algorhm ha furher refnes he global opmzaon sysem saus from he curren parcle fler sysem saus by usng he parcle swarm opmzaon mehod. The parcle swarm opmzaon mehod akes each parcle (saus) as an ndependen ndvdual by smulang he food fndng way of fsh school or nsecs, and can fnd he global opmal sysem saus under he muual cooperaon of all parcles. As a resul, PSO-PF can be used o search for he opmal lane model. In such an algorhm, he lane model s frsly defned. Each parcle represens a lane model and hus s known as a lane parcle. The algorhm can be dvded no wo sages. In he frs sage, he parcle fler s used for he predcon, measuremen and re-samplng of lane parcles. In he second sage, all lane parcles n he frs sage are aken as he orgnal parcles n he parcle swarm opmzaon mehod and he global opmal lane parcles can be obaned hrough calculaon by he parcle swarm opmzaon mehod. The remanng secons n hs paper are as follows: Secon 2 frsly nroduces he parcle fler and parcle swarm opmzaon mehod and hen specfes he new PSO-PF algorhm (PSO-PF). Secon 3 nroduces he lane model and how o use PSO-PF o complee he lane deecon and rackng. Secon 4 shows he expermens and resuls and he las secon presens he conclusons. 2. The PSO-PF Algorhm The PSO-PF algorhm hybrdzes a parcle fler and a parcle swarm opmzaon algorhm. The secon s gven over o llusrang he desgn of hs new algorhm Parcle Flers n Frsly, he dynamc sysem saus whn me s expressed as x R, = 1, 2,, N, wheren, he sysem saus conans he locaon, speed or range sze. The forecasng sysem of parcle fler s as follows: x = f ( x 1, w ), (1) n n n where f s he sysem ransfer marx from R R R, whch conans how o forecas he curren saus x based on he pas saus x 1. The forecased nose w s also added no he forecasng process, bu he nformaon s ndependen n me and s dsrbuon s no lmed. In he p observaon saus z R a me, z s a column vecor of p 1, conanng he observed saus resul. The observaon sysem s defned as follows: z = h ( x, v ), (2) n p p h s he observaon marx from R R R, he curren saus can be evaluaed as observaon resul z, and v s he nose durng observaon. The man purpose of parcle fler s o esmae he probably densy funcon wh he observed nformaon { z 1, z2,..., z}. In such case, D s he se of { z 1, z2,..., z}. Assume he poseror probably dsrbuon funcon (pdf) p( x 1 D 1) a me -1 s known, he poseror pdf p( x D ) can be deduced by Bayesan heorem. Ths process conans predcon and measuremen sages:

4 Sensors 2012, (1) Predcon: The poseror pdf p( x D 1) s propagaed a me sep usng he Chapman-Kolmogorov equaon: p( x D 1) = p( x x 1) p( x 1 D 1) dx 1, (3) wheren p( x x 1) s a Markov procedure. (2) Measuremen: The poseror pdf p( x D ) s compued usng he observaon vecor { z 1, z2,..., z} : p( z x ) p( x D 1 ) p( x D ) =, p( z D 1 ) (4) wheren P z ) s a normalzed em expressed as follows: ( D 1 p z D ) p( z x ) p( x D ) dx. (5) ( 1 = 1 The am of he PF algorhm s o recursvely esmae he poseror pdf p( x D ). Therefore, hs pdf s represened by a se of weghed parcles {( s, π ), = 1, 2,..., N}, where he weghs π p( z x = s ) are normalzed. The fnal sae oupu of he sysem s can be esmaed by he followng equaon: N s = π s (6) = 1 The basc parcle fler algorhm also needs o conduc he parcle re-samplng based on he parcle wegh n addon o he above observaon and forecas seps. The parcle wh hgher wegh π has more new parcles, and he oal number of new parcles s equal o ha of old parcles. Because of dfferen re-samplng mehods, many varable algorhms are proposed for he parcle fler. The CONDENSATION algorhm frsly proposed by Isard and Blake [22], whch s a parcle fler algorhm ha s easly mplemened and hus adoped n hs paper Parcle Swarm Opmzaon Algorhm Parcle Swarm Opmzaon (PSO) s a sochasc global opmzaon echnque whch has been shown o successfully opmze a wde range of connuous funcons [23,24]. The algorhm s based on a flockng of brds or a schoolng of fsh o look for food. PSO searches a space by adjusng he rajecores of ndvdual vecors, also called parcles, whch are concepualzed as movng pons n muldmensonal problem space wh wo assocaed vecors, poson vecor x k and velocy vecor v k, = 1, 2,, N for he curren evoluonary eraon k. The velocy of he ndvdual parcle n PSO s dynamcally adjused accordng o s own flyng experence and s companons flyng experence flyng n he search space. The former was ermed cognon-only model and he laer was ermed socal-only model [25]. Therefore, we ge: v k+ 1 = w v k + c1 rand ( x pbes xk ) + c2 rand ( x gbes xk )), (7) x = x + v k+ 1 k k, (8) where w s nera wegh facor (Hu and Eberha [26] suggesed usng 0.4~0.9), c 1 and c 2 are acceleraon consans. rand s random number beween 0 and 1. v s he velocy of parcle a k

5 Sensors 2012, eraon k. and x k s poson of parcle a eraon k, x gbes s he bes prevous poson among all he parcles. x pbes s he bes prevous poson of parcle 2.3. The PSO-PF Algorhm Ths secon wll nroduce he newly-proposed PSO-PF algorhm. To faclae he descrpon of algorhm, we rewre he parcle fler and parcle swarm opmzaon mehod wh a pseudo code program. Algorhm 1 shows a pseudo code program of a sandard SIR parcle fler. The algorhm conans he parameer seng and man loops. The parameer seng conans se P of N parcle sauses and se E of esmaed values of all parcles ha changes wh me. The man loop conans such seps as predcon, measuremen and resample of parcle saus, among whch he resample consss of he wegh calculaon of each parcle saus and he poson of new parcle n re-samplng based on he wegh value. Algorhm 2 shows he pseudo code algorhm of parcle swarm opmzaon mehod, whch also conans he parameer seng and man loops. As for he parameer seng, se P represens he se of N parcles, and se V represens he se of movemen speeds of all parcles n he k h eraon. Se P pbes s he se of recordng he opmal poson of each parcle, and se P gbes represens he weghed sum of P pbes of all curren parcles or he se of opmal posons so far. Algorhm 1. A pseudo-code for PF algorhm. Procedure ParcleFler() Var P[1 N]: Parcle se; E[1 N]: Esmae se along me; : me; begn = 0; Inalze(P); whle ( < T) do begn Predcon(P); Esmae(P, E[]); Resample(P, E[]); = + 1; end end Algorhm 2. A pseudo-code for PSO algorhm. Procedure ParcleSwarmOpmzaon() Var P[1 N]: parcle se; V[1 N]: velocy se along eraon; P pbes [1 N]: he bes parcle se along eraon; P gbes : he bes parcle of all parcle se along eraon; k: eraon;

6 Sensors 2012, Algorhm 2. Con. begn k = 0; Inalze(P); whle (k < K) do begn CompuePacleBes(P, P pbes [k]); CompueWeghBes(P pbes [k], P gbes [k]); CompueVelocys(V[k]); Updae(P); k = k + 1; end end As shown n Equaon (6), he parcle fler generally akes he weghed sum of all parcles or he parcle wh he bgges wegh value as he fnal oupu. However, boh oupus are approxmae opmal soluons or local opmal soluons. Thus n hs paper, we proposed a new PSO-PF algorhm by combnng he parcle fler wh he parcle swarm opmzaon mehod. Snce he parcle swarm opmzaon mehod s an algorhm o ge he opmal soluon, s used o furher refne he global opmal oupu from all parcle sauses afer he parcle fler re-samplng. Algorhm 3 shows he pseudo code of he PSO-PF algorhm, whch frsly nalzes all parcle sauses and hen uses he parcle fler o carry ou he predcon, measuremen and resample of parcle saus, and hen furher searches for he opmal saus from parcle sauses afer re-samplng by he parcle swarm opmzaon mehod. Laer, copes all parcle sauses afer he parcle fler re-samplng o he parcle swarm opmzaon mehod o sar he erave compung program. Snce he process of he parcle swarm opmzaon mehod doesn cause any mpac on all parcle sauses of parcle fler, can be conduced n parallel wh he algorhm. Algorhm 3. A pseudo-code for PSO-PF algorhm. Procedure PSOAlgorhmParcleFler() Var P[1 N]: Parcle se; E[1 N]: Esmae se along me; : me; begn (Parcle Fler) = 0; Inalze(P); whle ( < T) do begn Predcon(P); Esmae(P, E[]); Resample(P, E[]); k = 0; (Parcle Swarm Opmzaon ) Inalze(P = P);

7 Sensors 2012, Algorhm 3. Con. end end whle (k < K) do begn CompuePacleBes(P, P pbes [k]); CompueWeghBes(P pbes [k], P gbes [k]); CompueVelocys(V[k]); Updae(P ); k = k + 1; end = + 1; Fgure 1 shows he flow char of T vdeo frames connuously execued by he PSO-PF algorhm, n whch he upper par shows connuously npu vdeo frames and he lower he procedure flow processed by a dual core processor. A me, vdeo frame of frame() s npu and he core 1 frsly nalzes he parcle sae of he parcle fler, hen predcs, esmaes and re-samples parcles. Fnally, he parcle sae P s oupu o core 2 as he nalzaon parcle group P of PSO mehod and he parcle sae for he nex npu vdeo frame a me +1. When core 1 s processng he parcle fler for vdeo frame of frame(+1), core 2 s execung he PSO mehod of parcle group for vdeo frame of frame() and oupung he resul, P gbes. Therefore, PSO-PF algorhm can execue parallel processng. In addon, n he execuon, f fals o deec he lane, core 1 wll re-nalze he parcle sae of he parcle fler o connue deecng he nex npu vdeo frame. Fgure 1. The block dagram for PSO-PF algorhm. frame() frame(+1) frame(+t) Inalzed PF Inalzed PF Core 1 PF() P P +1 PF(+1) PF(+T) Core 2 P P +1 PSO() PSO(+1) PSO(+T-1) PF: Parcle Fler PSO: Parcle swarm opmzaon P gbes () P gbes (+1) P gbes (+T-1) 3. Lane Deecon and Trackng Usng he PSO-PF Ths secon llusraes how o use he proposed new PSO-PF algorhm (PSO-PF) o complee he lane deecon and rackng. Frsly, s mporan o defne he saus of each parcle ha represens

8 Sensors 2012, he lane model whch s also denfed as he lane parcle [21]. Wh gven parcles for N lanes, hen he PSO-PF algorhm s used o complee he predcon, measuremen and resample of lane parcle as well as search and oupu he opmal lane parcles, respecvely, as follows Lane Parcle General lane models represen he commonly-used sragh lne and curve lane models. Alhough he sragh lne lane model has a few parameers ha are easly calculaed, does no apply o he acual applcaon. Thus we adoped he curve lane model n hs paper. The ground lane model defned n hs paper s shown n Fgure 2. Fgure 2. Insallaon layou of lane models and cameras. β Z Y R C = 1 R α O β W X (X 0, Y 0, Z 0 ) Z L Gven he parcles for N lanes, he represenaon vecor of each lane parcle s shown as follows: T [ C, L, W, α, ], 1,..., N, x (9) = β = where C, L and W represen he horzonal curvaure, he laeral dsance from he vehcle cener o he rgh lane markng and wdh beween he lef and rgh lanes of he h lane parcle, respecvely. α and β denoe he pch angle and yaw angle of he camera. Relevan parameers are defned n he lef secon of Fgure 2. The rgh and lef drecons of he vehcle are shown n he X-axs whle he fron and back drecons are shown n he Z-axs, represenng he drecon of ravel of he vehcle. In he fgure, a par of double, hck doed arcs represens he poson of he lane markngs on he ground. The arc curvaure C s defned as he recprocal of he radus R of he correspondng arc. The rgh secon of Fgure 2 shows ha he es camera s nsalled n he vehcle below he rear vew mrror of he fron wndsheld whch s H cm above he ground. The camera faces he fron lane of he vehcle and s parallel o he vehcle body. There s a pch angle α wh he ground. Based on he assumpon of a fla ground, he pnhole camera s used o descrbe he relaonshp beween he camera coordnae and world coordnae. The pnhole camera model s based on he prncple of colneary, wheren each pon on he ground s projeced usng a sragh lne from he projecon cener o he mage plane. The orgn of he camera coordnae sysem s a he projecon cener of he locaon (X 0, Y 0, Z 0 ) wh respec o he ground coordnae sysem; he Z-axs of he camera frame s perpendcular o he mage plane. The roaon s represened usng Euler angles α (pch angle), β (yaw angle) and γ (roll angle)

9 Sensors 2012, whch defne a sequence of hree elemenary roaons around X, Y, Z-axs, respecvely. In addon, he drecon of he seerng wheel and he roll angle are no consdered n hs paper o smplfy he calculaon of he lane model. Fgure 3 shows he examples of he lane model and he correspondng projecon mage. I s assumed ha he vehcle s n he mddle of he lane, and Fgure 3(a c) represens he ground lane model layou of he lef bend (posve curvaure = 1e 5 ), sragh lne (curvaure s close o 0) and he rgh bend (negave curvaure = 1e 5 ), respecvely. The parallel doed lnes n he fgure represen he lef and rgh lane markngs (red lne for rgh lane markng, blue lne for lef lane markng) wh a 500 cm o 2,000 cm expresson range n fron of he vehcle. The nerval beween he wo crcles on he doed lne s 100 cm. Fgure 3(d f) shows he mages projeced by he lane models n Fgure 3(a c) hrough coordnae ransformaon, and he brgh spos n he mages represen he coordnaed pons a an nerval of 100 cm on he ground lane. Fgure 3. Examples of lane models and correspondng projecon mages lane model layous wh (a) posve curvaure, (b) zero curvaure, and (c) negave curvaure. (d f) are he correspondng projecon mages of lane models n (a c) (a) (b) (c) (d) (e) (f) 3.2. Predcon The purpose of parcle predcon s o forecas he curren saus x based on he pas saus x 1. In predcon, he forecased nose w s nroduced; herefore, he parcle predcon can be expressed as a smple dynamc moon equaon: x = Ax 1 + Bw, = 1,..., N, (10)

10 Sensors 2012, wheren A = dag ( I) and B = dag ([ α 1, α 2, α 3, α 4, α 5 ]). The forecased nose w can be expressed as a normalzed Gaussan nose vecor. Snce he sze of each parameer of he lane parcle vecor s dfferen, he forecased nose w vecor can be adjused hrough he B marx parameers when nroduced, so as o make suable o he sze of each parameer n he lane parcle. Usng Equaon (10), s easy o predc he nex saus for each lane parcle and o esmae he predcon resuls Measuremen The man purpose of he measuremen seps s o calculae each lane parcle and npu he machng degree of lane markng n he mages. As menoned n he above secons, he lane model represened by each lane parcle can be expressed as many coordnaes on he ground lane markngs (as shown n Fgure 3(a c)). These ground coordnae pons can be projeced on he mages hrough coordnae ransformaon (as shown n Fgure 3(d f)). Each pon n he mage receves a score based on he dsance o he neares lane markng and hese scores are used o compue he machng score of each lane parcle [22]. To complee he measuremen seps, we carred ou he lane markng deecon for npu mages. Snce he lane markng has noceable edges and characerscs such as hgh brghness, fxed wdh and drecvy, we no only deeced he edge resul bu also confrmed he poson of he lane markng based on he above characerscs. Fgure 4 shows an example, n whch Fgure 4(a) s he npu mage and Fgure 4(b) s he edge mage of he boom half npu mage. I can be obvously seen ha here also exs edges of oher panng lnes on he ground n addon o he lane edges. As for he lane markng deecon mehod whch only consders edges wll easly cause wrong deecon. Fgure 4(c) s he feaure mage of he npu mage whch s also he deecon resul of he edge mages based on prevous characerscs. For easer calculaon of he machng degree beween he lane parcle and lane markng, he dsance ransformaon s fnally used o ransform he lane markng mage no anoher dsance mage, as shown n Fgure 4(d). I s assumed ha he ground coordnae pon of he lane model of each lane parcle s projeced on he mage coordnae and can be expressed as ( u j, v j ), j = 1,, n, where n represens he number of coordnae pons as demonsraed by he brgh spos n Fgure 4(d). The wegh value of each lane parcle s proporonal o he measuremen resul. Therefore, he wegh value of each parcle can be expressed as follows: π = p( z x = s ) n 2 d( u, v ) j 1 j j p( Feaure ) exp (11) x = s =, = 1,..., N, 2 2nσ where d(u, v) represens he gray-scale value of coordnaes (u, v) of he dsance mage, ndcang ha he dsance from he coordnae pon o he neares lane markng. σ represens he sandard devaon o he Gaussan measuremen funcon. In he followng seps, he wegh value of each parcle s used o calculae he cdf funcon whch s adoped o deermne he poson of he lane parcle n re-samplng. Therefore, more new lane parcles can be obaned afer re-samplng near he poson of he lane parcle wh bgger wegh value; f no, less new lane parcles are obaned. As a resul, when he lane markngs n he npu mage are connuous, he wegh of he lane parcle near he lane s bg. Mos new lane parcles afer =

11 Sensors 2012, re-samplng gaher near he lane, hus he lane deecon and rackng can be compleed effecvely and quckly. However, when he lanes n he npu mage are dsconnuous, he wegh of he lane parcle near he new lane s small. However, here are sll a few correspondng new lane parcles afer re-samplng. Consequenly, afer re-adjusng he wegh of he lane parcle, he lane deecon and rackng can sll be compleed quckly and connuously n he followng npu mages. Fgure 4. Image measuremen examples of lane parcle, (a) npu mage, (b) edge mage, (c) lane markng characersc pon mage, and (d) dsance ransformaon mage and lane model coordnae pon projecon. (a) (b) (c) (d) 3.4. Lane Model Refned by PSO Based on he PSO-PF algorhm, he las sep uses he parcle swarm opmzaon mehod o complee he search for he oupu global opmal lane parcle. Therefore, he algorhm wll copy he lane parcle of he parcle flers as he nal saus parcle for he parcle swarm opmzaon mehod. Afer several eraons, he oupu global opmal lane parcle can be obaned, and wll be used n he follow-up lane devaon or ohers. Alhough he eraon of parcle swarm opmal mehod needs exra calculaon, he search process wh he parcle swarm opmzaon mehod doesn affec he operaon of parcle flers, hus can be operaed n parallel wh he algorhm. Fgure 5 shows he parcle dsrbuon map correspondng o he npu mage n Fgure 4. To show clearly, only hree parameers of lane parcle are shown n he fgure, namely horzonal curvaure C, laeral dsance L and lane markng wdh W. In hs example, he number of parcles s 20 and parameers w, c1, c2 are all se o 0.5, 1 and 1, respecvely. The blue dos n he fgure and green crcles represen he parcle dsrbuon of parcle flers and he correspondng oupu lane parcles respecvely. The mnmum dsance beween oupu lane parcle and lane s and he red asersk means he

12 Sensors 2012, global opmal lane parcle hrough he PSO-PF algorhm. The mnmum dsance beween he global opmal lane parcle and lane s 1.675, hus s clearly seen ha a smaller mnmum dsance can be obaned hrough PSO-PF algorhm (see Fgure 6 when usng 20 lane parcles). Fgure 5. Parcle Dsrbuon Example (N = 20). 200 Parcles The oupu of PF The ouou of PSO-PF Laeral dsance Xc (L) x Expermenal Resuls 1 0 Cv Wdh Horzonal curvaure (C ) Lane markng wdh (W ) In hs secon, we nroduce he expermenal resuls and envronmen. The equpmen used n he es ncluded one Inel 5 M GHz processor, 4 GB memory and Wndow 7 operang sysem whle he Malab sofware 2009 verson was used for developmen. The es road envronmen s composed manly of hghways, expressways and ordnary roads, ncludng sragh lne and curved lanes as well as up and down nerchanges. The es process also conans a connuous changng lane. Fgure 6 shows he comparson of measured resuls of oupu lane parcled beween he parcle fler and PSO-PF usng dfferen numbers of parcles as well as he dagram of he relaonshp beween he number of dfferen parcles of he parcle fler and he me consumpon. Each es resul value s he average value of 100 connuous npu mages. In Fgure 6, he blue crcle doed lne and green square doed lne represen he es resuls of he parcle fler and PSO-PF algorhm, respecvely. As shown n he fgure, f he parcle fler s used for lane deecon and rackng, he ncrease of parcle number wll resul n beer resuls, bu afer he parcles are ncreased connuously, he resuls wll end o be unchanged. However, when he parcle number ncreases, he requred compung me wll ncrease lnearly (see he red doed lne n he fgure). Therefore, f a hgher accuracy s needed, more parcles and more me wll be requred. On he conrary, fewer parcles mgh provde faser processng speed bu lower accuracy. The accuracy of PSO-PF algorhm for lane deecon and rackng s no affeced by he parcle number, and dfferen parcle numbers are adoped, hus he accuracy of PSO-PF algorhm s hgher han ha of parcle fler. The use of PSO-PF algorhm can ge he opmzed resul wh a small number of parcles. Snce he number of parcle used s small, he compung me and memory can also be saved. Alhough he PSO-PF algorhm needs addonal

13 Sensors 2012, compung me, s parcle fler and parcle swarm opmzaon mehod can be calculaed n parallel. When he sysem s mplemened, he acual me won be ncreased. Consequenly, he use of PSO-PF algorhm can realze he opmzaon resul by usng a few parcles, whch can effecvely mprove he effcency of parcle fler for lane deecon and rackng. Accordng o expermens, he execuon me of parcle fler under Malab envronmen s abou 0.7 sec/frame when usng 50 lane parcles whle ha of PSO mehod s abou 0.5 sec/frame. Therefore, under parallel processng for PSO-PF, he bgger value 0.7 sec/frame s aken, by whch can be known ha he me requred by PSO-PF algorhm s abou he execuon me of parcle fler. In praccal operaon, PSO-PF algorhm s mplemened on a dual-core embedded sysem wh C language and runnng speed of abou 15 fps. Fgure 6. Comparson beween parcle fler and PSO-PF by usng dfferen parcle numbers and mnmum dsances, and relaonshp dagram of parcle fler beween dfferen parcle numbers and me. 2.5 PF PSO-PF Tmes(sec.) 2 Measures or seconds The number of parcles The oupu lane parcle resuls of he successvely npu 800 frames (.e., he npu vdeo mages) are shown n Fgures 7 and 8, where he number of lane parcles s 50. Fgure 7 shows he comparson on he laeral dsance L of he oupu lane parcles beween he parcle fler and PSO-PF algorhm as he lane deecon and rackng (un: cm). The vehcle lne crossng warnng sgnal s also shown n Fgure 7. When he lef wheel of he vehcle crosses he lef lane markng, he lef wheel lne cross warnng sgnal s gven, as shown n he blue square wave below Fgure 7. When he rgh wheel of he vehcle crosses he rgh lane markng, he rgh wheel lne cross warnng sgnal s gven as shown n he green square wave below Fgure 7. I can be observed n Fgure 7 ha four lane changes occurred n he es. The frs wo are changes from curren lanes o lef lanes, wh he lef wheel lne cross warnng sgnal occurrng frsly followed by he rgh wheel lne cross warnng sgnal. The las wo are changes from he curren lanes o he rgh lanes. Therefore, he rgh wheel lne cross warnng sgnal occurs frsly followed by he lef wheel lne cross warnng sgnal. The duraon of he warnng sgnal s relaed o he exchange speed. In addon, accordng o Fgure 7, he oupu resuls of parcle fler s smooher han ha of PSO-PF algorhm, whch s manly because he oupu lane parcles of

14 Sensors 2012, parcle fler adop he weghed sum of all lane parcles, whle hose of PSO-PF algorhm adop he global opmal lane parcles. Furhermore, he camera shake also causes obvous changes n he oupu lane parcle resuls of PSO-PF algorhm due o he hgh-speed movemen of he vehcle. Fgure 7. Comparson beween parcle fler and PSO-PF algorhm on he laeral dsance of oupu lane parcles. 400 pf pf+pso ldw Laeral dsance Frames Fgure 8. Comparson beween Parcle Fler and PSO-PF algorhm on he mnmum dsance of oupu lane parcles. 25 pf pf+pso 20 Mnmal dsances Frames Fgure 8 shows he comparson of he parcle fler and PSO-PF algorhm on he mnmum dsance beween oupu lane parcle and lane markng (un: pxel). I can be clearly found ha he mnmum dsance of he oupu lane parcle by usng he PSO-PF algorhm s smaller han ha of parcle fler, especally when he lane markng s dsconnuous or changes. As for frames n he fgure wh frame numbers less han 125, can be found ha when he vehcle goes no he hghway from nerchange, he lane markng s shorened and dsconnuous, and he frame poson changes around 90 and 125 once for each me. The resuls show ha he PSO-PF algorhm can effecvely

15 Sensors 2012, mprove he resuls of he parcle fler, alhough n some crcumsances, here s sll lle dfference beween our proposed mehod and parcle fler, as frame posons around 265 and 380 show n he fgure. Fgure 9 shows he connuous mage of a lane change process n whch vehcle changes from he curren lane o he lef lane (whch corresponds o he second lane change shown n Fgure 7). In he fgure, an mage s dsplayed every 5 frames. The blue lne n he mage represens he lef lane markng and he green lne represens he rgh lane markng, as shown n Fgure 9(a c). When he lef wheel of he vehcle crosses he lane markng, he blue color on he lef edge changes o red and he lef wheel lne cross warnng sgnal s gven, as shown n Fgure 9(d f). When he vehcle cener crosses he lef lane markng, he orgnal lef lane edge becomes he rgh lane markng of he new lane, and he rgh wheel of he vehcle crosses he rgh lane markng. Therefore, he rgh lane markng n he mage changes from green o red and he rgh wheel lne cross warnng sgnal s gven, as shown n Fgure 9(g,h). When he vehcle complees he lane change, he rgh lane markng n he mage urns back o green and he rgh wheel lne cross warnng sgnal s cancelled, as shown n Fgure 9( l). The orgnal move fles showng he es resuls can be downloaded from hp:// Fgure 9. Connuous Images of a lane change example. (a) (b) (c) (d) (e) (f) (g) (h) () (j) (k) (l)

16 Sensors 2012, Conclusons In hs paper, we have proposed a new algorhm o mprove he effcency of he parcle fler, whch s known as he parcle swarm opmzaon parcle fler (PSO-PF). The algorhm combnes he parcle fler and he parcle swarm opmzaon mehod and has he followng properes: (1) he algorhm uses he parcle swarm opmzaon mehod o furher search for he global opmal oupu sysem saus from among he sysem sauses of he parcle fler, herefore, he opmal soluon o a problem can be obaned; (2) snce each parcle fler execued ncludes he predcon, measuremen and resample of he sysem saus, he compuaon s rapd. Whle he parcle swarm opmzaon mehod s an algorhm based on eraon, hus he compung process requres more me, he PSO-PF algorhm uses he parcle fler whch can compue quckly o search for he local opmal soluon or approxmae opmal soluon, and hen uses he parcle swarm opmzaon mehod o search for he global opmal soluon, henceforh, hs resuls n a beer effcency; (3) boh he parcle fler and parcle swarm opmzaon mehod are soluons based on he parcles, hus hey are easy o negrae and mplemen; (4) snce he parcle fler and parcle swarm opmzaon mehod don affec each oher, hey can be compued n parallel o ncrease he overall calculaon speed; (5) as shown n he expermenal resuls, he parcle number of he parcle fler s proporonal o he calculaon speed and also o he oupu accuracy, hus he parcle fler can ge beer oupu bu requre more me f more parcles are used. On he conrary, f fewer parcles are used, less me wll be needed for calculaon bu worse resuls wll be obaned. However, he PSO-PF algorhm can acheve beer effcency whle usng fewer parcles. Fnally, we used he PSO-PF algorhm o carry ou he lane deecon and rackng mmedaely and hen compared s resuls wh hose of he parcle fler. The expermenal resuls verfy ha he mehod can effecvely mprove he resuls of he parcle fler for lane deecon and rackng. References 1. Mccall, J.C.; Trved, M.M. Vdeo-based lane esmaon and rackng for drver asssance: survey, sysem, and evaluaon. IEEE Trans. Inellg. Transp. Sys. 2006, 7, Hsao, P.-Y.; Yeh, C.-W.; Huang, S.-S.; Fu, L.-C. A porable vson-based real-me lane deparure warnng sysem: day and ngh. IEEE Trans. Vehc. Technol. 2009, 58, Clanon, J.M.; Bevly, D.M.; Hodel, S. A low-cos soluon for an negraed mulsensor lane deparure warnng sysem. IEEE Trans. Inell. Transp. Sys. 2009, 10, Schuber, R.; Schulze, K.; Wanelk, G. Suaon assessmen for auomac lane-change maneuvers. IEEE Trans. Inell. Transpor. Sys. 2010, 11, Angkrakul, P.; Terashma, R.; Waka, T. On he use of sochasc drver behavor model n lane deparure warnng. IEEE Trans. Inell. Transp. Sys. 2011, 12, McCall, J.C.; Trved, M.M. Vsual conex capure and analyss for drver aenon monorng. In Proceedngs of he IEEE Conference on Inellgen Transporaon Sysems, Washngon, DC, USA, 3 6 Ocober 2004, pp

17 Sensors 2012, Kreucher, C.; Lakshmanan, S.; Kluge, K. A drver warnng sysem based on he LOIS lane deecon algorhm. In Proceedngs of he IEEE Conference on Inellgen Vehcles, Sugar, Germany, Ocober 1998, pp Taylor, C.; Kosecka, J.; Blas, R.; Malk, J. A comparave sudy of vson-based laeral conrol sraeges for auonomous hghway drvbg. In. J. Robo. Res. 1999, 18, Aufrere, R.; Chapus, R.; Chausse, F. A Model-drven approach for real-me road recognon. Mach. Vs. Appl. 2001, 13, Nedevsch, S.; Schmd, R.; Graf, T.; Danescu, R.; Frenu, D.; Mara, T.; Onga, F.; Pocol, C. 3D lane deecon sysem based on sereovson. In Proceedngs of he IEEE Conference on Inellgen Transporaon Sysems, Washngon, DC, USA, 3 6 Ocober 2004; pp Labayrade, R.; Doure, J.; Auber, D. A Mul-model lane deecor ha handles road sngulares. In Proceedngs of he IEEE Conference on Inellgen Transporaon Sysems, Torono, ON, Canada, Sepember 2006; pp Franke, U.; Loose, H.; Knoeppel, C. Lane recognon on counry roads. In Proceedngs of IEEE Inellgen Vehcles Symposum, Isanbul, Turkey, June 2007; pp Lm, K.H.; Seng, K.P.; Ngo, A.C.L.; Ang, L.-M. Real-me mplemenaon of vson-based lane deecon and rackng. In Procedngs of he Inernaonal Conference on Inellgen Human-Machne Sysems and Cybernecs, Hangzhou, Chna, Augus 2009; pp Wang, J.; Cheng, Y.; Xe, J.; Ln, H. Model-based lane deecon and lane followng for nellgen vehcles. In Proceedngs of he Inernaonal Conference on Inellgen Human-Machne Sysems and Cybernecs, Nanjng, Chna, Augus 2010; pp Chen, C.-Y.; Chen, C.-H.; Da, Z.-X. An evoluonary compuaon approach for lane deecon and rackng. Advan. Sc. Le. 2012, 9, Souhall, B.; Taylor, C.J. Sochasc road shape esmaon. In Proceedngs of he IEEE Inernaonal Conference on Compuer Vson, Vancouver, BC, Canada, 7 14 July 2001; pp Aposoloff, N.; Zelnsky, A. Robus vson based lane rackng usng mulple cues and parcle flerng. In Proceedngs of he IEEE Inellgen Vehcles Symposum, Columbus, OH, USA, 9 11 June 2003; pp Macek, K.; Wllams, B.; Kolsk, S.; Segwar, R. A lane deecon vson module for drver asssance. In Proceedngs of he IEEE/APS Conference on Mecharoncs and Robocs, Aachen, Germany, Sepember 2004; pp Smuda, P.; Schweger, R.; Neumann, H.; Rer, W. Mulple cue daa fuson wh parcle flers for road course deecon n vson sysems. In Proceedngs of he IEEE Inellgen Vehcle Symposum Tokyo, Japan, June 2006; pp Km, Z. Robus lane deecon and rackng n challengng scenaros. IEEE Trans. Inell. Transp. Sys. 2008, 9, Danescu, R.; Nedevsch, S. Probablsc lane rackng n dffcul road scenaros usng sereovson. IEEE Trans. Inell. Transp. Sys. 2009, 10, Isard, M.; Blake, A. Condensaon: Condonal densy propagaon for vsual rackng. In. J. Compu. Vs. 1998, 29, 5 28.

18 Sensors 2012, Kennedy, J.; Eberhar, R.C. Parcle swarm opmzaon. In Proceedngs of he IEEE Inernaonal Conference on Neural Neworks, Pscaaway, NJ, USA, 27 November 1 December 1995; pp Eberhar, R.C.; Sh, Y. Trackng and opmzng dynamc sysems wh parcle swarms. In Proceedngs of he Congress on Evoluonary Compuaon, Seoul, Korea, May 2001; pp Hu, X.; Eberhar, R. C. Trackng dynamc sysems wh PSO: Where s he cheese? In Proceedngs of he Workshop on Parcle Swarm Opmzaon, Indanapols, IN, USA, 6 7 Aprl Parsopoulos, K.E.; Vrahas, M.N. Recen approaches o global opmzaon problems hrough parcle swarm opmzaon. Neur. Compu. 2002, 1, by he auhors; lcensee MDPI, Basel, Swzerland. Ths arcle s an open access arcle dsrbued under he erms and condons of he Creave Commons Arbuon lcense (hp://creavecommons.org/lcenses/by/3.0/).

Fall 2010 Graduate Course on Dynamic Learning

Fall 2010 Graduate Course on Dynamic Learning Fall 200 Graduae Course on Dynamc Learnng Chaper 4: Parcle Flers Sepember 27, 200 Byoung-Tak Zhang School of Compuer Scence and Engneerng & Cognve Scence and Bran Scence Programs Seoul aonal Unversy hp://b.snu.ac.kr/~bzhang/

More information

On One Analytic Method of. Constructing Program Controls

On One Analytic Method of. Constructing Program Controls Appled Mahemacal Scences, Vol. 9, 05, no. 8, 409-407 HIKARI Ld, www.m-hkar.com hp://dx.do.org/0.988/ams.05.54349 On One Analyc Mehod of Consrucng Program Conrols A. N. Kvko, S. V. Chsyakov and Yu. E. Balyna

More information

Variants of Pegasos. December 11, 2009

Variants of Pegasos. December 11, 2009 Inroducon Varans of Pegasos SooWoong Ryu bshboy@sanford.edu December, 009 Youngsoo Cho yc344@sanford.edu Developng a new SVM algorhm s ongong research opc. Among many exng SVM algorhms, we wll focus on

More information

Dynamic Team Decision Theory. EECS 558 Project Shrutivandana Sharma and David Shuman December 10, 2005

Dynamic Team Decision Theory. EECS 558 Project Shrutivandana Sharma and David Shuman December 10, 2005 Dynamc Team Decson Theory EECS 558 Proec Shruvandana Sharma and Davd Shuman December 0, 005 Oulne Inroducon o Team Decson Theory Decomposon of he Dynamc Team Decson Problem Equvalence of Sac and Dynamc

More information

CS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 4

CS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 4 CS434a/54a: Paern Recognon Prof. Olga Veksler Lecure 4 Oulne Normal Random Varable Properes Dscrmnan funcons Why Normal Random Varables? Analycally racable Works well when observaon comes form a corruped

More information

Solution in semi infinite diffusion couples (error function analysis)

Solution in semi infinite diffusion couples (error function analysis) Soluon n sem nfne dffuson couples (error funcon analyss) Le us consder now he sem nfne dffuson couple of wo blocks wh concenraon of and I means ha, n a A- bnary sysem, s bondng beween wo blocks made of

More information

WiH Wei He

WiH Wei He Sysem Idenfcaon of onlnear Sae-Space Space Baery odels WH We He wehe@calce.umd.edu Advsor: Dr. Chaochao Chen Deparmen of echancal Engneerng Unversy of aryland, College Par 1 Unversy of aryland Bacground

More information

Cubic Bezier Homotopy Function for Solving Exponential Equations

Cubic Bezier Homotopy Function for Solving Exponential Equations Penerb Journal of Advanced Research n Compung and Applcaons ISSN (onlne: 46-97 Vol. 4, No.. Pages -8, 6 omoopy Funcon for Solvng Eponenal Equaons S. S. Raml *,,. Mohamad Nor,a, N. S. Saharzan,b and M.

More information

doi: info:doi/ /

doi: info:doi/ / do: nfo:do/0.063/.322393 nernaonal Conference on Power Conrol and Opmzaon, Bal, ndonesa, -3, June 2009 A COLOR FEATURES-BASED METHOD FOR OBJECT TRACKNG EMPLOYNG A PARTCLE FLTER ALGORTHM Bud Sugand, Hyoungseop

More information

Particle Filter Based Robot Self-localization Using RGBD Cues and Wheel Odometry Measurements Enyang Gao1, a*, Zhaohua Chen1 and Qizhuhui Gao1

Particle Filter Based Robot Self-localization Using RGBD Cues and Wheel Odometry Measurements Enyang Gao1, a*, Zhaohua Chen1 and Qizhuhui Gao1 6h Inernaonal Conference on Elecronc, Mechancal, Informaon and Managemen (EMIM 206) Parcle Fler Based Robo Self-localzaon Usng RGBD Cues and Wheel Odomery Measuremens Enyang Gao, a*, Zhaohua Chen and Qzhuhu

More information

V.Abramov - FURTHER ANALYSIS OF CONFIDENCE INTERVALS FOR LARGE CLIENT/SERVER COMPUTER NETWORKS

V.Abramov - FURTHER ANALYSIS OF CONFIDENCE INTERVALS FOR LARGE CLIENT/SERVER COMPUTER NETWORKS R&RATA # Vol.) 8, March FURTHER AALYSIS OF COFIDECE ITERVALS FOR LARGE CLIET/SERVER COMPUTER ETWORKS Vyacheslav Abramov School of Mahemacal Scences, Monash Unversy, Buldng 8, Level 4, Clayon Campus, Wellngon

More information

THE PREDICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS

THE PREDICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS THE PREICTION OF COMPETITIVE ENVIRONMENT IN BUSINESS INTROUCTION The wo dmensonal paral dfferenal equaons of second order can be used for he smulaon of compeve envronmen n busness The arcle presens he

More information

Robustness Experiments with Two Variance Components

Robustness Experiments with Two Variance Components Naonal Insue of Sandards and Technology (NIST) Informaon Technology Laboraory (ITL) Sascal Engneerng Dvson (SED) Robusness Expermens wh Two Varance Componens by Ana Ivelsse Avlés avles@ns.gov Conference

More information

HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD

HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD Journal of Appled Mahemacs and Compuaonal Mechancs 3, (), 45-5 HEAT CONDUCTION PROBLEM IN A TWO-LAYERED HOLLOW CYLINDER BY USING THE GREEN S FUNCTION METHOD Sansław Kukla, Urszula Sedlecka Insue of Mahemacs,

More information

In the complete model, these slopes are ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL. (! i+1 -! i ) + [(!") i+1,q - [(!

In the complete model, these slopes are ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL. (! i+1 -! i ) + [(!) i+1,q - [(! ANALYSIS OF VARIANCE FOR THE COMPLETE TWO-WAY MODEL The frs hng o es n wo-way ANOVA: Is here neracon? "No neracon" means: The man effecs model would f. Ths n urn means: In he neracon plo (wh A on he horzonal

More information

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore. Ths documen s downloaded from DR-NTU, Nanyang Technologcal Unversy Lbrary, Sngapore. Tle A smplfed verb machng algorhm for word paron n vsual speech processng( Acceped verson ) Auhor(s) Foo, Say We; Yong,

More information

MANY real-world applications (e.g. production

MANY real-world applications (e.g. production Barebones Parcle Swarm for Ineger Programmng Problems Mahamed G. H. Omran, Andres Engelbrech and Ayed Salman Absrac The performance of wo recen varans of Parcle Swarm Opmzaon (PSO) when appled o Ineger

More information

GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS. Youngwoo Ahn and Kitae Kim

GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS. Youngwoo Ahn and Kitae Kim Korean J. Mah. 19 (2011), No. 3, pp. 263 272 GENERATING CERTAIN QUINTIC IRREDUCIBLE POLYNOMIALS OVER FINITE FIELDS Youngwoo Ahn and Kae Km Absrac. In he paper [1], an explc correspondence beween ceran

More information

Detection of Waving Hands from Images Using Time Series of Intensity Values

Detection of Waving Hands from Images Using Time Series of Intensity Values Deecon of Wavng Hands from Images Usng Tme eres of Inensy Values Koa IRIE, Kazunor UMEDA Chuo Unversy, Tokyo, Japan re@sensor.mech.chuo-u.ac.jp, umeda@mech.chuo-u.ac.jp Absrac Ths paper proposes a mehod

More information

CHAPTER 10: LINEAR DISCRIMINATION

CHAPTER 10: LINEAR DISCRIMINATION CHAPER : LINEAR DISCRIMINAION Dscrmnan-based Classfcaon 3 In classfcaon h K classes (C,C,, C k ) We defned dscrmnan funcon g j (), j=,,,k hen gven an es eample, e chose (predced) s class label as C f g

More information

New M-Estimator Objective Function. in Simultaneous Equations Model. (A Comparative Study)

New M-Estimator Objective Function. in Simultaneous Equations Model. (A Comparative Study) Inernaonal Mahemacal Forum, Vol. 8, 3, no., 7 - HIKARI Ld, www.m-hkar.com hp://dx.do.org/.988/mf.3.3488 New M-Esmaor Objecve Funcon n Smulaneous Equaons Model (A Comparave Sudy) Ahmed H. Youssef Professor

More information

Kernel-Based Bayesian Filtering for Object Tracking

Kernel-Based Bayesian Filtering for Object Tracking Kernel-Based Bayesan Flerng for Objec Trackng Bohyung Han Yng Zhu Dorn Comancu Larry Davs Dep. of Compuer Scence Real-Tme Vson and Modelng Inegraed Daa and Sysems Unversy of Maryland Semens Corporae Research

More information

Lecture 6: Learning for Control (Generalised Linear Regression)

Lecture 6: Learning for Control (Generalised Linear Regression) Lecure 6: Learnng for Conrol (Generalsed Lnear Regresson) Conens: Lnear Mehods for Regresson Leas Squares, Gauss Markov heorem Recursve Leas Squares Lecure 6: RLSC - Prof. Sehu Vjayakumar Lnear Regresson

More information

( ) () we define the interaction representation by the unitary transformation () = ()

( ) () we define the interaction representation by the unitary transformation () = () Hgher Order Perurbaon Theory Mchael Fowler 3/7/6 The neracon Represenaon Recall ha n he frs par of hs course sequence, we dscussed he chrödnger and Hesenberg represenaons of quanum mechancs here n he chrödnger

More information

Lecture VI Regression

Lecture VI Regression Lecure VI Regresson (Lnear Mehods for Regresson) Conens: Lnear Mehods for Regresson Leas Squares, Gauss Markov heorem Recursve Leas Squares Lecure VI: MLSC - Dr. Sehu Vjayakumar Lnear Regresson Model M

More information

Filtrage particulaire et suivi multi-pistes Carine Hue Jean-Pierre Le Cadre and Patrick Pérez

Filtrage particulaire et suivi multi-pistes Carine Hue Jean-Pierre Le Cadre and Patrick Pérez Chaînes de Markov cachées e flrage parculare 2-22 anver 2002 Flrage parculare e suv mul-pses Carne Hue Jean-Perre Le Cadre and Parck Pérez Conex Applcaons: Sgnal processng: arge rackng bearngs-onl rackng

More information

Including the ordinary differential of distance with time as velocity makes a system of ordinary differential equations.

Including the ordinary differential of distance with time as velocity makes a system of ordinary differential equations. Soluons o Ordnary Derenal Equaons An ordnary derenal equaon has only one ndependen varable. A sysem o ordnary derenal equaons consss o several derenal equaons each wh he same ndependen varable. An eample

More information

Object Tracking Based on Visual Attention Model and Particle Filter

Object Tracking Based on Visual Attention Model and Particle Filter Inernaonal Journal of Informaon Technology Vol. No. 9 25 Objec Trackng Based on Vsual Aenon Model and Parcle Fler Long-Fe Zhang, Yuan-Da Cao 2, Mng-Je Zhang 3, Y-Zhuo Wang 4 School of Compuer Scence and

More information

Linear Response Theory: The connection between QFT and experiments

Linear Response Theory: The connection between QFT and experiments Phys540.nb 39 3 Lnear Response Theory: The connecon beween QFT and expermens 3.1. Basc conceps and deas Q: ow do we measure he conducvy of a meal? A: we frs nroduce a weak elecrc feld E, and hen measure

More information

Sampling Procedure of the Sum of two Binary Markov Process Realizations

Sampling Procedure of the Sum of two Binary Markov Process Realizations Samplng Procedure of he Sum of wo Bnary Markov Process Realzaons YURY GORITSKIY Dep. of Mahemacal Modelng of Moscow Power Insue (Techncal Unversy), Moscow, RUSSIA, E-mal: gorsky@yandex.ru VLADIMIR KAZAKOV

More information

Computing Relevance, Similarity: The Vector Space Model

Computing Relevance, Similarity: The Vector Space Model Compung Relevance, Smlary: The Vecor Space Model Based on Larson and Hears s sldes a UC-Bereley hp://.sms.bereley.edu/courses/s0/f00/ aabase Managemen Sysems, R. Ramarshnan ocumen Vecors v ocumens are

More information

Performance Analysis for a Network having Standby Redundant Unit with Waiting in Repair

Performance Analysis for a Network having Standby Redundant Unit with Waiting in Repair TECHNI Inernaonal Journal of Compung Scence Communcaon Technologes VOL.5 NO. July 22 (ISSN 974-3375 erformance nalyss for a Nework havng Sby edundan Un wh ang n epar Jendra Sngh 2 abns orwal 2 Deparmen

More information

5th International Conference on Advanced Design and Manufacturing Engineering (ICADME 2015)

5th International Conference on Advanced Design and Manufacturing Engineering (ICADME 2015) 5h Inernaonal onference on Advanced Desgn and Manufacurng Engneerng (IADME 5 The Falure Rae Expermenal Sudy of Specal N Machne Tool hunshan He, a, *, La Pan,b and Bng Hu 3,c,,3 ollege of Mechancal and

More information

John Geweke a and Gianni Amisano b a Departments of Economics and Statistics, University of Iowa, USA b European Central Bank, Frankfurt, Germany

John Geweke a and Gianni Amisano b a Departments of Economics and Statistics, University of Iowa, USA b European Central Bank, Frankfurt, Germany Herarchcal Markov Normal Mxure models wh Applcaons o Fnancal Asse Reurns Appendx: Proofs of Theorems and Condonal Poseror Dsrbuons John Geweke a and Gann Amsano b a Deparmens of Economcs and Sascs, Unversy

More information

Short-Term Load Forecasting Using PSO-Based Phase Space Neural Networks

Short-Term Load Forecasting Using PSO-Based Phase Space Neural Networks Proceedngs of he 5h WSEAS In. Conf. on SIMULATION, MODELING AND OPTIMIZATION, Corfu, Greece, Augus 7-9, 005 (pp78-83) Shor-Term Load Forecasng Usng PSO-Based Phase Space Neural Neworks Jang Chuanwen, Fang

More information

A Novel Object Detection Method Using Gaussian Mixture Codebook Model of RGB-D Information

A Novel Object Detection Method Using Gaussian Mixture Codebook Model of RGB-D Information A Novel Objec Deecon Mehod Usng Gaussan Mxure Codebook Model of RGB-D Informaon Lujang LIU 1, Gaopeng ZHAO *,1, Yumng BO 1 1 School of Auomaon, Nanjng Unversy of Scence and Technology, Nanjng, Jangsu 10094,

More information

Approximate Analytic Solution of (2+1) - Dimensional Zakharov-Kuznetsov(Zk) Equations Using Homotopy

Approximate Analytic Solution of (2+1) - Dimensional Zakharov-Kuznetsov(Zk) Equations Using Homotopy Arcle Inernaonal Journal of Modern Mahemacal Scences, 4, (): - Inernaonal Journal of Modern Mahemacal Scences Journal homepage: www.modernscenfcpress.com/journals/jmms.aspx ISSN: 66-86X Florda, USA Approxmae

More information

Robust and Accurate Cancer Classification with Gene Expression Profiling

Robust and Accurate Cancer Classification with Gene Expression Profiling Robus and Accurae Cancer Classfcaon wh Gene Expresson Proflng (Compuaonal ysems Bology, 2005) Auhor: Hafeng L, Keshu Zhang, ao Jang Oulne Background LDA (lnear dscrmnan analyss) and small sample sze problem

More information

Math 128b Project. Jude Yuen

Math 128b Project. Jude Yuen Mah 8b Proec Jude Yuen . Inroducon Le { Z } be a sequence of observed ndependen vecor varables. If he elemens of Z have a on normal dsrbuon hen { Z } has a mean vecor Z and a varancecovarance marx z. Geomercally

More information

Lecture 18: The Laplace Transform (See Sections and 14.7 in Boas)

Lecture 18: The Laplace Transform (See Sections and 14.7 in Boas) Lecure 8: The Lalace Transform (See Secons 88- and 47 n Boas) Recall ha our bg-cure goal s he analyss of he dfferenal equaon, ax bx cx F, where we emloy varous exansons for he drvng funcon F deendng on

More information

Clustering (Bishop ch 9)

Clustering (Bishop ch 9) Cluserng (Bshop ch 9) Reference: Daa Mnng by Margare Dunham (a slde source) 1 Cluserng Cluserng s unsupervsed learnng, here are no class labels Wan o fnd groups of smlar nsances Ofen use a dsance measure

More information

FTCS Solution to the Heat Equation

FTCS Solution to the Heat Equation FTCS Soluon o he Hea Equaon ME 448/548 Noes Gerald Reckenwald Porland Sae Unversy Deparmen of Mechancal Engneerng gerry@pdxedu ME 448/548: FTCS Soluon o he Hea Equaon Overvew Use he forward fne d erence

More information

Time-interval analysis of β decay. V. Horvat and J. C. Hardy

Time-interval analysis of β decay. V. Horvat and J. C. Hardy Tme-nerval analyss of β decay V. Horva and J. C. Hardy Work on he even analyss of β decay [1] connued and resuled n he developmen of a novel mehod of bea-decay me-nerval analyss ha produces hghly accurae

More information

Algorithm Research on Moving Object Detection of Surveillance Video Sequence *

Algorithm Research on Moving Object Detection of Surveillance Video Sequence * Opcs and Phooncs Journal 03 3 308-3 do:0.436/opj.03.3b07 Publshed Onlne June 03 (hp://www.scrp.org/journal/opj) Algorhm Research on Movng Objec Deecon of Survellance Vdeo Sequence * Kuhe Yang Zhmng Ca

More information

Computer Robot Vision Conference 2010

Computer Robot Vision Conference 2010 School of Compuer Scence McGll Unversy Compuer Robo Vson Conference 2010 Ioanns Rekles Fundamenal Problems In Robocs How o Go From A o B? (Pah Plannng) Wha does he world looks lke? (mappng) sense from

More information

EEL 6266 Power System Operation and Control. Chapter 5 Unit Commitment

EEL 6266 Power System Operation and Control. Chapter 5 Unit Commitment EEL 6266 Power Sysem Operaon and Conrol Chaper 5 Un Commmen Dynamc programmng chef advanage over enumeraon schemes s he reducon n he dmensonaly of he problem n a src prory order scheme, here are only N

More information

Outline. Probabilistic Model Learning. Probabilistic Model Learning. Probabilistic Model for Time-series Data: Hidden Markov Model

Outline. Probabilistic Model Learning. Probabilistic Model Learning. Probabilistic Model for Time-series Data: Hidden Markov Model Probablsc Model for Tme-seres Daa: Hdden Markov Model Hrosh Mamsuka Bonformacs Cener Kyoo Unversy Oulne Three Problems for probablsc models n machne learnng. Compung lkelhood 2. Learnng 3. Parsng (predcon

More information

Using Fuzzy Pattern Recognition to Detect Unknown Malicious Executables Code

Using Fuzzy Pattern Recognition to Detect Unknown Malicious Executables Code Usng Fuzzy Paern Recognon o Deec Unknown Malcous Execuables Code Boyun Zhang,, Janpng Yn, and Jngbo Hao School of Compuer Scence, Naonal Unversy of Defense Technology, Changsha 40073, Chna hnxzby@yahoo.com.cn

More information

Attribute Reduction Algorithm Based on Discernibility Matrix with Algebraic Method GAO Jing1,a, Ma Hui1, Han Zhidong2,b

Attribute Reduction Algorithm Based on Discernibility Matrix with Algebraic Method GAO Jing1,a, Ma Hui1, Han Zhidong2,b Inernaonal Indusral Informacs and Compuer Engneerng Conference (IIICEC 05) Arbue educon Algorhm Based on Dscernbly Marx wh Algebrac Mehod GAO Jng,a, Ma Hu, Han Zhdong,b Informaon School, Capal Unversy

More information

Motion in Two Dimensions

Motion in Two Dimensions Phys 1 Chaper 4 Moon n Two Dmensons adzyubenko@csub.edu hp://www.csub.edu/~adzyubenko 005, 014 A. Dzyubenko 004 Brooks/Cole 1 Dsplacemen as a Vecor The poson of an objec s descrbed by s poson ecor, r The

More information

Chapter 6: AC Circuits

Chapter 6: AC Circuits Chaper 6: AC Crcus Chaper 6: Oulne Phasors and he AC Seady Sae AC Crcus A sable, lnear crcu operang n he seady sae wh snusodal excaon (.e., snusodal seady sae. Complee response forced response naural response.

More information

Population Density Particle Swarm Optimized Improved Multi-robot Cooperative Localization Algorithm

Population Density Particle Swarm Optimized Improved Multi-robot Cooperative Localization Algorithm WSES TSCTIOS on CICUITS and SYSTEMS Mengchu Tan, Yumng o Zhmn Chen, Panlong Wu, Cong Yue Populaon Densy Parcle Swarm Opmzed Improved Mul-robo Cooperave Localzaon lgorhm Mengchu TI, Yumng O, Zhmn CHE*,

More information

( t) Outline of program: BGC1: Survival and event history analysis Oslo, March-May Recapitulation. The additive regression model

( t) Outline of program: BGC1: Survival and event history analysis Oslo, March-May Recapitulation. The additive regression model BGC1: Survval and even hsory analyss Oslo, March-May 212 Monday May 7h and Tuesday May 8h The addve regresson model Ørnulf Borgan Deparmen of Mahemacs Unversy of Oslo Oulne of program: Recapulaon Counng

More information

On computing differential transform of nonlinear non-autonomous functions and its applications

On computing differential transform of nonlinear non-autonomous functions and its applications On compung dfferenal ransform of nonlnear non-auonomous funcons and s applcaons Essam. R. El-Zahar, and Abdelhalm Ebad Deparmen of Mahemacs, Faculy of Scences and Humanes, Prnce Saam Bn Abdulazz Unversy,

More information

Bernoulli process with 282 ky periodicity is detected in the R-N reversals of the earth s magnetic field

Bernoulli process with 282 ky periodicity is detected in the R-N reversals of the earth s magnetic field Submed o: Suden Essay Awards n Magnecs Bernoull process wh 8 ky perodcy s deeced n he R-N reversals of he earh s magnec feld Jozsef Gara Deparmen of Earh Scences Florda Inernaonal Unversy Unversy Park,

More information

ISSN MIT Publications

ISSN MIT Publications MIT Inernaonal Journal of Elecrcal and Insrumenaon Engneerng Vol. 1, No. 2, Aug 2011, pp 93-98 93 ISSN 2230-7656 MIT Publcaons A New Approach for Solvng Economc Load Dspach Problem Ansh Ahmad Dep. of Elecrcal

More information

Anisotropic Behaviors and Its Application on Sheet Metal Stamping Processes

Anisotropic Behaviors and Its Application on Sheet Metal Stamping Processes Ansoropc Behavors and Is Applcaon on Shee Meal Sampng Processes Welong Hu ETA-Engneerng Technology Assocaes, Inc. 33 E. Maple oad, Sue 00 Troy, MI 48083 USA 48-79-300 whu@ea.com Jeanne He ETA-Engneerng

More information

II. Light is a Ray (Geometrical Optics)

II. Light is a Ray (Geometrical Optics) II Lgh s a Ray (Geomercal Opcs) IIB Reflecon and Refracon Hero s Prncple of Leas Dsance Law of Reflecon Hero of Aleandra, who lved n he 2 nd cenury BC, posulaed he followng prncple: Prncple of Leas Dsance:

More information

Relative controllability of nonlinear systems with delays in control

Relative controllability of nonlinear systems with delays in control Relave conrollably o nonlnear sysems wh delays n conrol Jerzy Klamka Insue o Conrol Engneerng, Slesan Techncal Unversy, 44- Glwce, Poland. phone/ax : 48 32 37227, {jklamka}@a.polsl.glwce.pl Keywor: Conrollably.

More information

Comparison of Differences between Power Means 1

Comparison of Differences between Power Means 1 In. Journal of Mah. Analyss, Vol. 7, 203, no., 5-55 Comparson of Dfferences beween Power Means Chang-An Tan, Guanghua Sh and Fe Zuo College of Mahemacs and Informaon Scence Henan Normal Unversy, 453007,

More information

A Bayesian algorithm for tracking multiple moving objects in outdoor surveillance video

A Bayesian algorithm for tracking multiple moving objects in outdoor surveillance video A Bayesan algorhm for racng mulple movng obecs n oudoor survellance vdeo Manunah Narayana Unversy of Kansas Lawrence, Kansas manu@u.edu Absrac Relable racng of mulple movng obecs n vdes an neresng challenge,

More information

Polymerization Technology Laboratory Course

Polymerization Technology Laboratory Course Prakkum Polymer Scence/Polymersaonsechnk Versuch Resdence Tme Dsrbuon Polymerzaon Technology Laboraory Course Resdence Tme Dsrbuon of Chemcal Reacors If molecules or elemens of a flud are akng dfferen

More information

Li An-Ping. Beijing , P.R.China

Li An-Ping. Beijing , P.R.China A New Type of Cpher: DICING_csb L An-Png Bejng 100085, P.R.Chna apl0001@sna.com Absrac: In hs paper, we wll propose a new ype of cpher named DICING_csb, whch s derved from our prevous sream cpher DICING.

More information

Existence and Uniqueness Results for Random Impulsive Integro-Differential Equation

Existence and Uniqueness Results for Random Impulsive Integro-Differential Equation Global Journal of Pure and Appled Mahemacs. ISSN 973-768 Volume 4, Number 6 (8), pp. 89-87 Research Inda Publcaons hp://www.rpublcaon.com Exsence and Unqueness Resuls for Random Impulsve Inegro-Dfferenal

More information

Bayes rule for a classification problem INF Discriminant functions for the normal density. Euclidean distance. Mahalanobis distance

Bayes rule for a classification problem INF Discriminant functions for the normal density. Euclidean distance. Mahalanobis distance INF 43 3.. Repeon Anne Solberg (anne@f.uo.no Bayes rule for a classfcaon problem Suppose we have J, =,...J classes. s he class label for a pxel, and x s he observed feaure vecor. We can use Bayes rule

More information

An introduction to Support Vector Machine

An introduction to Support Vector Machine An nroducon o Suppor Vecor Machne 報告者 : 黃立德 References: Smon Haykn, "Neural Neworks: a comprehensve foundaon, second edon, 999, Chaper 2,6 Nello Chrsann, John Shawe-Tayer, An Inroducon o Suppor Vecor Machnes,

More information

Notes on the stability of dynamic systems and the use of Eigen Values.

Notes on the stability of dynamic systems and the use of Eigen Values. Noes on he sabl of dnamc ssems and he use of Egen Values. Source: Macro II course noes, Dr. Davd Bessler s Tme Seres course noes, zarads (999) Ineremporal Macroeconomcs chaper 4 & Techncal ppend, and Hamlon

More information

Ordinary Differential Equations in Neuroscience with Matlab examples. Aim 1- Gain understanding of how to set up and solve ODE s

Ordinary Differential Equations in Neuroscience with Matlab examples. Aim 1- Gain understanding of how to set up and solve ODE s Ordnary Dfferenal Equaons n Neuroscence wh Malab eamples. Am - Gan undersandng of how o se up and solve ODE s Am Undersand how o se up an solve a smple eample of he Hebb rule n D Our goal a end of class

More information

Appendix to Online Clustering with Experts

Appendix to Online Clustering with Experts A Appendx o Onlne Cluserng wh Expers Furher dscusson of expermens. Here we furher dscuss expermenal resuls repored n he paper. Ineresngly, we observe ha OCE (and n parcular Learn- ) racks he bes exper

More information

Chapter 6 DETECTION AND ESTIMATION: Model of digital communication system. Fundamental issues in digital communications are

Chapter 6 DETECTION AND ESTIMATION: Model of digital communication system. Fundamental issues in digital communications are Chaper 6 DEECIO AD EIMAIO: Fundamenal ssues n dgal communcaons are. Deecon and. Esmaon Deecon heory: I deals wh he desgn and evaluaon of decson makng processor ha observes he receved sgnal and guesses

More information

PARTICLE FILTER BASED VEHICLE TRACKING APPROACH WITH IMPROVED RESAMPLING STAGE

PARTICLE FILTER BASED VEHICLE TRACKING APPROACH WITH IMPROVED RESAMPLING STAGE ISS: 0976-910(OLIE) ICTACT JOURAL O IMAGE AD VIDEO PROCESSIG, FEBRUARY 014, VOLUME: 04, ISSUE: 03 PARTICLE FILTER BASED VEHICLE TRACKIG APPROACH WITH IMPROVED RESAMPLIG STAGE We Leong Khong 1, We Yeang

More information

The Analysis of the Thickness-predictive Model Based on the SVM Xiu-ming Zhao1,a,Yan Wang2,band Zhimin Bi3,c

The Analysis of the Thickness-predictive Model Based on the SVM Xiu-ming Zhao1,a,Yan Wang2,band Zhimin Bi3,c h Naonal Conference on Elecrcal, Elecroncs and Compuer Engneerng (NCEECE The Analyss of he Thcknesspredcve Model Based on he SVM Xumng Zhao,a,Yan Wang,band Zhmn B,c School of Conrol Scence and Engneerng,

More information

Department of Economics University of Toronto

Department of Economics University of Toronto Deparmen of Economcs Unversy of Torono ECO408F M.A. Economercs Lecure Noes on Heeroskedascy Heeroskedascy o Ths lecure nvolves lookng a modfcaons we need o make o deal wh he regresson model when some of

More information

Graduate Macroeconomics 2 Problem set 5. - Solutions

Graduate Macroeconomics 2 Problem set 5. - Solutions Graduae Macroeconomcs 2 Problem se. - Soluons Queson 1 To answer hs queson we need he frms frs order condons and he equaon ha deermnes he number of frms n equlbrum. The frms frs order condons are: F K

More information

ELASTIC MODULUS ESTIMATION OF CHOPPED CARBON FIBER TAPE REINFORCED THERMOPLASTICS USING THE MONTE CARLO SIMULATION

ELASTIC MODULUS ESTIMATION OF CHOPPED CARBON FIBER TAPE REINFORCED THERMOPLASTICS USING THE MONTE CARLO SIMULATION THE 19 TH INTERNATIONAL ONFERENE ON OMPOSITE MATERIALS ELASTI MODULUS ESTIMATION OF HOPPED ARBON FIBER TAPE REINFORED THERMOPLASTIS USING THE MONTE ARLO SIMULATION Y. Sao 1*, J. Takahash 1, T. Masuo 1,

More information

[ ] 2. [ ]3 + (Δx i + Δx i 1 ) / 2. Δx i-1 Δx i Δx i+1. TPG4160 Reservoir Simulation 2018 Lecture note 3. page 1 of 5

[ ] 2. [ ]3 + (Δx i + Δx i 1 ) / 2. Δx i-1 Δx i Δx i+1. TPG4160 Reservoir Simulation 2018 Lecture note 3. page 1 of 5 TPG460 Reservor Smulaon 08 page of 5 DISCRETIZATIO OF THE FOW EQUATIOS As we already have seen, fne dfference appromaons of he paral dervaves appearng n he flow equaons may be obaned from Taylor seres

More information

Effect of Resampling Steepness on Particle Filtering Performance in Visual Tracking

Effect of Resampling Steepness on Particle Filtering Performance in Visual Tracking 102 The Inernaonal Arab Journal of Informaon Technology, Vol. 10, No. 1, January 2013 Effec of Resamplng Seepness on Parcle Flerng Performance n Vsual Trackng Zahdul Islam, Ch-Mn Oh, and Chl-Woo Lee School

More information

Single-loop System Reliability-Based Design & Topology Optimization (SRBDO/SRBTO): A Matrix-based System Reliability (MSR) Method

Single-loop System Reliability-Based Design & Topology Optimization (SRBDO/SRBTO): A Matrix-based System Reliability (MSR) Method 10 h US Naonal Congress on Compuaonal Mechancs Columbus, Oho 16-19, 2009 Sngle-loop Sysem Relably-Based Desgn & Topology Opmzaon (SRBDO/SRBTO): A Marx-based Sysem Relably (MSR) Mehod Tam Nguyen, Junho

More information

A Fuzzy Model for the Multiobjective Emergency Facility Location Problem with A-Distance

A Fuzzy Model for the Multiobjective Emergency Facility Location Problem with A-Distance The Open Cybernecs and Sysemcs Journal, 007, 1, 1-7 1 A Fuzzy Model for he Mulobecve Emergency Facly Locaon Problem wh A-Dsance T. Uno *, H. Kaagr and K. Kao Deparmen of Arfcal Complex Sysems Engneerng,

More information

e-journal Reliability: Theory& Applications No 2 (Vol.2) Vyacheslav Abramov

e-journal Reliability: Theory& Applications No 2 (Vol.2) Vyacheslav Abramov June 7 e-ournal Relably: Theory& Applcaons No (Vol. CONFIDENCE INTERVALS ASSOCIATED WITH PERFORMANCE ANALYSIS OF SYMMETRIC LARGE CLOSED CLIENT/SERVER COMPUTER NETWORKS Absrac Vyacheslav Abramov School

More information

Neural Networks-Based Time Series Prediction Using Long and Short Term Dependence in the Learning Process

Neural Networks-Based Time Series Prediction Using Long and Short Term Dependence in the Learning Process Neural Neworks-Based Tme Seres Predcon Usng Long and Shor Term Dependence n he Learnng Process J. Puchea, D. Paño and B. Kuchen, Absrac In hs work a feedforward neural neworksbased nonlnear auoregresson

More information

CamShift Guided Particle Filter for Visual Tracking

CamShift Guided Particle Filter for Visual Tracking CamShf Guded Parcle Fler for Vsual Trackng Zhaowen Wang, Xaokang Yang, Y Xu and Songyu Yu Insue of Image Communcaon and Informaon Processng Shangha Jao Tong Unversy, Shangha, PRC 200240 E-mal: {whereaswll,xkyang,

More information

Lecture 11 SVM cont

Lecture 11 SVM cont Lecure SVM con. 0 008 Wha we have done so far We have esalshed ha we wan o fnd a lnear decson oundary whose margn s he larges We know how o measure he margn of a lnear decson oundary Tha s: he mnmum geomerc

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 5 Lecure 9 Hamlonan Equaons of Moon (Chaper 8) Wha We Dd Las Tme Consruced Hamlonan formalsm H ( q, p, ) = q p L( q, q, ) H p = q H q = p H = L Equvalen o Lagrangan formalsm Smpler, bu

More information

Reactive Methods to Solve the Berth AllocationProblem with Stochastic Arrival and Handling Times

Reactive Methods to Solve the Berth AllocationProblem with Stochastic Arrival and Handling Times Reacve Mehods o Solve he Berh AllocaonProblem wh Sochasc Arrval and Handlng Tmes Nsh Umang* Mchel Berlare* * TRANSP-OR, Ecole Polyechnque Fédérale de Lausanne Frs Workshop on Large Scale Opmzaon November

More information

UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 2017 EXAMINATION

UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 2017 EXAMINATION INTERNATIONAL TRADE T. J. KEHOE UNIVERSITAT AUTÒNOMA DE BARCELONA MARCH 27 EXAMINATION Please answer wo of he hree quesons. You can consul class noes, workng papers, and arcles whle you are workng on he

More information

FI 3103 Quantum Physics

FI 3103 Quantum Physics /9/4 FI 33 Quanum Physcs Aleander A. Iskandar Physcs of Magnesm and Phooncs Research Grou Insu Teknolog Bandung Basc Conces n Quanum Physcs Probably and Eecaon Value Hesenberg Uncerany Prncle Wave Funcon

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 5 Lecure 9 Hamlonan Equaons of Moon (Chaper 8) Wha We Dd Las Tme Consruced Hamlonan formalsm Hqp (,,) = qp Lqq (,,) H p = q H q = p H L = Equvalen o Lagrangan formalsm Smpler, bu wce as

More information

Machine Learning Linear Regression

Machine Learning Linear Regression Machne Learnng Lnear Regresson Lesson 3 Lnear Regresson Bascs of Regresson Leas Squares esmaon Polynomal Regresson Bass funcons Regresson model Regularzed Regresson Sascal Regresson Mamum Lkelhood (ML)

More information

Chapter Lagrangian Interpolation

Chapter Lagrangian Interpolation Chaper 5.4 agrangan Inerpolaon Afer readng hs chaper you should be able o:. dere agrangan mehod of nerpolaon. sole problems usng agrangan mehod of nerpolaon and. use agrangan nerpolans o fnd deraes and

More information

Introduction to Boosting

Introduction to Boosting Inroducon o Boosng Cynha Rudn PACM, Prnceon Unversy Advsors Ingrd Daubeches and Rober Schapre Say you have a daabase of news arcles, +, +, -, -, +, +, -, -, +, +, -, -, +, +, -, + where arcles are labeled

More information

PHYS 1443 Section 001 Lecture #4

PHYS 1443 Section 001 Lecture #4 PHYS 1443 Secon 001 Lecure #4 Monda, June 5, 006 Moon n Two Dmensons Moon under consan acceleraon Projecle Moon Mamum ranges and heghs Reerence Frames and relae moon Newon s Laws o Moon Force Newon s Law

More information

Volatility Interpolation

Volatility Interpolation Volaly Inerpolaon Prelmnary Verson March 00 Jesper Andreasen and Bran Huge Danse Mares, Copenhagen wan.daddy@danseban.com brno@danseban.com Elecronc copy avalable a: hp://ssrn.com/absrac=69497 Inro Local

More information

Real-time Quantum State Estimation Based on Continuous Weak Measurement and Compressed Sensing

Real-time Quantum State Estimation Based on Continuous Weak Measurement and Compressed Sensing Proceedngs of he Inernaonal MulConference of Engneers and Compuer Scenss 018 Vol II IMECS 018, March 14-16, 018, Hong Kong Real-me Quanum Sae Esmaon Based on Connuous Weak Measuremen and Compressed Sensng

More information

Single and Multiple Object Tracking Using a Multi-Feature Joint Sparse Representation

Single and Multiple Object Tracking Using a Multi-Feature Joint Sparse Representation Sngle and Mulple Objec Trackng Usng a Mul-Feaure Jon Sparse Represenaon Wemng Hu, We L, and Xaoqn Zhang (Naonal Laboraory of Paern Recognon, Insue of Auomaon, Chnese Academy of Scences, Bejng 100190) {wmhu,

More information

Hongbin Dong Computer Science and Technology College Harbin Engineering University Harbin, China

Hongbin Dong Computer Science and Technology College Harbin Engineering University Harbin, China Hongbn Dong Compuer Scence and Harbn Engneerng Unversy Harbn, Chna donghongbn@hrbeu.edu.cn Xue Yang Compuer Scence and Harbn Engneerng Unversy Harbn, Chna yangxue@hrbeu.edu.cn Xuyang Teng Compuer Scence

More information

Pendulum Dynamics. = Ft tangential direction (2) radial direction (1)

Pendulum Dynamics. = Ft tangential direction (2) radial direction (1) Pendulum Dynams Consder a smple pendulum wh a massless arm of lengh L and a pon mass, m, a he end of he arm. Assumng ha he fron n he sysem s proporonal o he negave of he angenal veloy, Newon s seond law

More information

Hidden Markov Models

Hidden Markov Models 11-755 Machne Learnng for Sgnal Processng Hdden Markov Models Class 15. 12 Oc 2010 1 Admnsrva HW2 due Tuesday Is everyone on he projecs page? Where are your projec proposals? 2 Recap: Wha s an HMM Probablsc

More information

ABSTRACT KEYWORDS. Bonus-malus systems, frequency component, severity component. 1. INTRODUCTION

ABSTRACT KEYWORDS. Bonus-malus systems, frequency component, severity component. 1. INTRODUCTION EERAIED BU-MAU YTEM ITH A FREQUECY AD A EVERITY CMET A IDIVIDUA BAI I AUTMBIE IURACE* BY RAHIM MAHMUDVAD AD HEI HAAI ABTRACT Frangos and Vronos (2001) proposed an opmal bonus-malus sysems wh a frequency

More information

Introduction ( Week 1-2) Course introduction A brief introduction to molecular biology A brief introduction to sequence comparison Part I: Algorithms

Introduction ( Week 1-2) Course introduction A brief introduction to molecular biology A brief introduction to sequence comparison Part I: Algorithms Course organzaon Inroducon Wee -2) Course nroducon A bref nroducon o molecular bology A bref nroducon o sequence comparson Par I: Algorhms for Sequence Analyss Wee 3-8) Chaper -3, Models and heores» Probably

More information