Ant Colony Optimization 209

Size: px
Start display at page:

Download "Ant Colony Optimization 209"

Transcription

1 Ant Colony Optmzaton 9 14 x Ant Colony Optmzaton Benlan Xu, Jhong Zhu and Qnlan Chen Changshu Insttute of Technology Chna 1. Introducton Swarm ntellgence s a relatvely novel approach to problem solvng that taes nspraton from the socal behavors of nsects and of other anmals. In partcular, ants have nspred a number of methods and technques among whch the most studed and the most successful one s the ant colony optmzaton. Ant colony optmzaton (ACO) algorthm, a novel populaton-based and meta-heurstc approach, was recently proposed by Dorgo et al. to solve several dscrete optmzaton problems (Dorgo, 1996, 1997). The general ACO algorthm mmcs the way real ants fnd the shortest route between a food source and ther nest. The ants communcate wth one another by means of pheromone trals and exchange nformaton ndrectly about whch path should be followed. Paths wth hgher pheromone levels wll more lely be chosen and thus renforced later, whle the pheromone ntensty of paths that are not chosen s decreased by evaporaton. Ths form of ndrect communcaton s nown as stgmergy, and provdes the ant colony shortest-path fndng capabltes. The frst algorthm followng the prncples of the ACO meta-heurstc s the Ant System (AS) (Dorgo,1996), where ants teratvely construct solutons and add pheromone to the paths correspondng to these solutons. Path selecton s a stochastc procedure based on two parameters, the pheromone and heurstc values, whch wll be detaled n the followng secton n ths chapter. The pheromone value gves an ndcaton of the number of ants that chose the tral recently, whle the heurstc value s problem-dependent and t has dfferent forms for dfferent cases. Due to the fact that the general ACO can be easly extended to deal wth other optmzaton problems, ts several varants has been proposed as well, such as Ant Colony System (Dorgo,1997), ran-based Ant System (Bullnhemer,1999), and Eltst Ant System (Dorgo,1996). And the above varants of ACO have been appled to a varety of dfferent problems, such as vehcle routng (Montemann,5), schedulng (Blum,5), and travellng salesman problem (Stützle,). Recently, ants have also entered the data mnng doman, addressng both the clusterng (Kanade,7), and classfcaton tas (Martens et al.,7). Ths chapter wll focus on another applcaton of ACO to trac ntaton n the target tracng feld. To the best of our nowledge, there are few reports on the trac ntaton usng the ACO. But n the real world, t s oerved that there s a case n whch almost all ants are nclned to gather around the food sources n the form of lne or curve. Fg. 1 shows the evoluton process of ants searchng for foods. Intally, all ants are dstrbuted randomly

2 1 New Advances n Machne Learnng n the plane as n Fg.1 (a), and a few hours later we fnd that most of ants gather together around the food sources as shown n Fg.1 (b). Tang nspraton from such phenomenon, we may regard these lnear or curvy food sources as tentatve tracs to be ntalzed, and the correspondng ant model s establshed from the optmal aspect to solve the problem of multple trac ntaton. food source 1 food source 1 food source food source (a) Intal dstrbuton of ants (b) The dstrbuton of ants a few hours later Fg. 1. The evoluton process of ant search for foods The remander of ths chapter s structured as follows. Frst, n secton, the wdely used ant system and ts successors are ntroduced. Secton 3 gves the new applcaton of ACO to the trac ntaton problem, and the system of ants of dfferent tass s modeled to concde wth the problem. The performance comparson of ACO-based technques for trac ntaton s carred out and analyszed n Secton 4. Fnally, some conclusons are drawn.. Ant System and Its Drect Successors.1 Ant System Intally, three dfferent versons of AS were developed (Dorgo et al., 1991), namely antdensty, ant-quantty, and ant-cycle. In the ant-densty and ant-quantty versons the ants updated the pheromone drectly after a move from one cty to an adacent cty, whle n the ant-cycle verson the pheromone update was only done after all the ants had constructed the tours and the amount of pheromone deposted by each ant was set to be a functon of the tour qualty. The two man phases of the AS algorthm consttute the ants soluton constructon and the pheromone update. In AS, a good way to ntalze the pheromone trals s to set them to a value slghtly hgher than the expected amount of pheromone deposted by the ants n one teraton. The reason for ths choce s that f the ntal pheromone values are too low, then the search s qucly based by the frst tours generated by the ants, whch n general leads toward the exploraton of nferor zones of the search space. On the other sde, f the ntal pheromone values are too hgh, then many teratons are lost watng untl pheromone evaporaton reduces enough pheromone values, so that pheromone added by ants can start to bas the search. Tour Constructon In AS, m (artfcal) ants ncrementally buld a tour of the TSP. Intally, ants are put on randomly chosen ctes. At each constructon step, ant apples a probablstc acton choce

3 Ant Colony Optmzaton 11 rule, called random proportonal rule, to decde whch cty to vst next. In partcular, the probablty wth whch ant, located at cty, chooses to go to cty s [ ] [ ] p, f N, [ ] [ ] ln l where 1/ d s a heurstc value that s computed n advance, and are two parameters whch determne the relatve mportance of the pheromone tral and the heurstc nformaton, and N s the set of ctes that ant has not vsted so far. By ths probablstc rule, the probablty of choosng the arc (, ) may ncrease wth the bgger value of the assocated pheromone tral and of the heurstc nformaton value. The role of the parameters and s descrbed as below. If, the closest ctes are more lely to be selected: ths corresponds to a classc stochastc greedy algorthm. If, t means that the pheromone s used alone, wthout any heurstc bas. Ths generally leads to rather poor results and, n partcular, for values of 1 t leads to earler stagnaton stuaton, that s, a stuaton n whch all the ants follow the same path and construct the same tour, whch, n general, s strongly suboptmal. Each ant mantans a memory whch records the ctes already vsted. And moreover, ths memory s used to defne the feasble neghbourhood N n the constructon rule gven by equaton (1). In addton, such a memory allows ant both to compute the length of the tour T t generated and to retrace the path to depost pheromone for upcomng global pheromone update. Concernng soluton constructon, there are two dfferent ways of mplementng t: parallel and sequental soluton constructon. In the parallel mplementaton, at each constructon step all ants move from ther current cty to the next one, whle n the sequental mplementaton an ant bulds a complete tour before the next one starts to buld another one. In the AS case, both choces for the mplementaton of the tour constructon are equvalent n the sense that they do not sgnfcantly nfluence the algorthm s behavour. Update of Pheromone Trals After all the ants have constructed ther tours, the pheromone trals are updated. Ths s done by frst lowerng the pheromone value on all arcs by a factor, and then addng an amount of pheromone on the arcs the ants have crossed n ther tours. Pheromone evaporaton s mplemented by the followng law (1 ), (, ) L () where 1 s the pheromone evaporaton rate. The parameter s used to avod unlmted accumulaton of the pheromone trals and t enables the algorthm to forget bad decsons prevously taen. In fact, f an arc s not chosen by the ants, ts assocated pheromone value decreases exponentally wth the number of teratons. After evaporaton, all ants depost pheromone on the arcs they have crossed n ther tour: l (1)

4 1 New Advances n Machne Learnng where m 1 (1 ), (, ) L (3) s the amount of pheromone ant deposts on the arcs t has vsted. It s defned as follows: 1/ C f arc (, ) belongs to T (4) otherw where C, the length of the tour T travelled by ant, s computed as the sum of the lengths of the arcs belongng to T. By means of equaton (4), the shorter an ant s tour s, the more pheromone the arcs belongng to ths tour receve. In general, arcs that are used by many ants and whch are part of short tours, receve more pheromone and are, therefore, more lely to be chosen by ants n the followng teratons of the algorthm... Eltst Ant System The eltst strategy for Ant System (EAS) (Dorgo,1996) s, n prncple, to provde a strong addtonal renforcement to the arcs belongng to the best tour found snce the start of the algorthm. Note that ths addtonal feedbac to the best-so-far tour s another example of a daemon acton of the ACO meta-heurstcs. Update of Pheromone Trals The addtonal renforcement of tour T s acheved by addng a quantty e / C to ts arcs, where e s a parameter that defnes the weght gven to the best-so-far tour T, and C s ts length. Thus, equaton (3) for the pheromone depost becomes where s defned as n equaton (4) and m e 1 (5) s defned as follows: 1/ C f arc (, ) belongs to T (6) otherw Note that n EAS the pheromone evaporaton s mplemented as n AS..3. Ran-Based Ant System Another mprovement over AS (Bullnhemer,1999) s the ran-based verson of AS ( AS ran ). In AS ran each ant deposts an amount of pheromone that decreases wth ts ran. Addtonally, as n EAS, the best-so-far ant always receves the largest amount of pheromone n each teraton. Update of Pheromone Trals Before updatng the pheromone trals, the ants are sorted by ncreasng tour length and the quantty of pheromone an ant deposts s weghted accordng to the ran of the ant. In each

5 Ant Colony Optmzaton 13 teraton, assume that total W best-raned ants are consdered, and only the ( W 1) bestraned ants and the ant that produced the best-so-far tour are allowed to depost pheromone. The best-so-far tour gves the strongest feedbac wth weght w ; the r th best r ant of the current teraton contrbutes to pheromone updatng wth the value 1/ C multpled by a weght gven by,w r. Thus, the AS ran pheromone update rule s r r where 1/ C and 1/ C. W 1 r ( W r) w r 1 (7).4 Max- Mn Ant System Max-Mn Ant System (MMAS) (St ü tzle & Hoos, ) ntroduces some man modfcatons wth respect to AS. Frst, t strongly explots the best tours found: only ether the teraton-best ant, that s, the ant that produced the best tour n the current teraton, or the best-so-far ant s allowed to depost pheromone. Unfortunately, such a strategy may lead to a stagnaton stuaton n whch all ants follow the same tour, because of the excessve growth of pheromone trals on arcs of a good, although suboptmal, tour. To counteract ths effect, a second modfcaton ntroduced by MMAS s that t lmts the possble range of pheromone tral values to the nterval [ mn, ]. Second, the pheromone trals are ntalzed to the upper pheromone tral lmt, whch, together wth a small pheromone evaporaton rate, ncreases the exploraton of tours at the start of the search. Fnally, n MMAS, pheromone trals are rentalzed each tme the system approaches stagnaton or when no mproved tour has been generated for a certan number of consecutve teratons. Update of Pheromone Trals After all ants have constructed a tour, pheromones are updated by applyng evaporaton as n AS, followed by the depost of new pheromone as follows: best, (8) best best where 1/ C. The ant whch s allowed to add pheromone may be ether the best- best best b so-far, n whch case 1/ C, or the teraton-best, n whch case 1/ C, b where C s the length of the teraton-best tour. In general, n MMAS mplementatons both the teraton-best and the best-so-far update rules are used n an alternate way. Obvously, the choce of the relatve frequency wth whch the two pheromone update rules are appled has an nfluence on how greedy the search s: When pheromone updates are always performed by the best-so-far ant, the search focuses very qucly around T, whereas when t s the teraton-best ant that updates pheromones, then the number of arcs that receve pheromone s larger and the search s less drected. Pheromone Tral Lmts In MMAS, lower and upper lmts mn and on the possble pheromone values on any arc are mposed n order to avod earler searchng stagnaton. In partcular, the mposed

6 14 New Advances n Machne Learnng pheromone tral lmts have the effect of lmtng the probablty when an ant s n cty to the nterval [ mn, ] p of selectng a cty p p p 1. Only p p, wth mn when an ant has ust one sngle possble choce for the next cty, that s N 1, we have p mn p 1. It s easy to show that, n the long run, the upper pheromone tral lmt on any arc s * * bounded by 1/ C, where C s the length of the optmal tour. Based on ths result, MMAS uses an estmate of ths value, 1/ C, to defne : each tme a new best-so-far tour s found, the value of s updated. The lower pheromone tral lmt s set to / mn, where s a parameter (Stützle & Hoos, ). Pheromone Tral Intalzaton and Re-ntalzaton At the start of the algorthm, the ntal pheromone trals are set to an estmate of the upper pheromone tral lmt. Ths way of ntalzng the pheromone trals, n combnaton wth a small pheromone evaporaton parameter, causes a slow ncrease n the relatve dfference n the pheromone tral levels, so that the ntal search phase of MMAS s very exploratve. Note that, n MMAS, pheromone trals are occasonally re-ntalzed. Pheromone tral rentalzaton s typcally trggered when the algorthm approaches the stagnaton behavour or f for a gven number of algorthm teratons no mproved tour s found. 3. ACO for Trac Intaton of Bearngs-only mult-target tracng 3.1 Problem Presentaton Bearngs-only mult-target tracng (BO-MTT) (Nardone, 1984 ; Dogancay, 4, 5) n a tatc system can be descrbed as: gven a tme hstory of nose-corrupted bearng measurements from two oervers, the obectve s to obtan optmum estmaton of the postons, veloctes and acceleratons of all targets. Generally, the whole process of target tracng ncludes trac ntaton, trac mantenance and trac deleton. To the best of our nowledge, however, many reported lterature manly focused on the trac mantenance,.e. target tracng, wthout consderng the trac ntaton process, after the moton of each target s modelled. Actually, trac ntaton plays an mportant role n evaluatng the performance of suequent target tracng, and mproperly ntated tracs may ether lead to target loss or the ncrease of consumpton of lmted resources. In the case of mult-sensor-mult-target BOT, for nstance, two-sensor-two-target BOT at a gven scan, four Lne of Sghts (LOSs) are avalable alone to determne whch LOS belongs to some target of nterest. Usually, such a problem can also be dealt wth the general trac ntaton technques wdely used n the radar tracng feld through ntersectng these LOSs to obtan a group of canddates of true targets poston ponts. However, such an operaton wll result n some ntersectons ncludng both the true target postons and the vrtual target poston called ghost, as shown n Fg.. These ghosts, n fact, do not belong to any target (denoted by poston ponts 3 and 4). Due to ths fact, the orgn uncertanty of obtaned poston canddates should be dscrmnated and ths ssue forms the topc of ths secton. In addton, such a problem becomes harder to handle n the presence of clutter.

7 Ant Colony Optmzaton 15 Y trac of target trac of target sensor 1 sensor O Fg.. The generated ghosts n case of two-sensor-two-target BOT X 3. Motve In the mage detecton feld, the Hough transform (H-T) has been recognzed as a robust technque for lne or curve detecton and also have been largely appled by scentfc communty (Bhattacharya, ; Shapro, 5). The basc dea of H-T s to transform a pont ( x, y) n the Cartesan coordnate system onto a curve n the (, ) parameter space, whch s formulated as x cos y sn (9) where s the dstance from the lne through ( x, y ) to the orgn, and s the angle to the normal wth the x axs. The angle vares from to 18, whle the may be ether postve or negatve. So, t s oerved that, f a set of ponts n the Cartesan coordnate le on the same lne, all curves each correspondng to a pont must ntersect at a same pont denoted by (, ) n the parameter space. Inspred by ths phenomenon, the H-T technque can be utlzed to ntalze the trac of target whch maes a unform rectlnear moton. 3.3 Soluton to Mult-Target Trac Intaton by ACO In ths secton, we wll nvestgate the problem of mult-target trac ntaton. Frst, a obectve functon s presented to descrbe the property of the mult-target trac ntaton. Second, a novel ACO algorthm, called dfferent tass of ants, s modelled to ntate the tracs of nterest. As noted before, f there are n curves n the parameter space, at most C n ntersectons are obtaned n general. However, n a real tracng scenaro, these curves wll not strctly ntersect the pont but several ponts dstrbuted n the parameter space due to the exstence

8 16 New Advances n Machne Learnng of measurement error. Even so, these ponts are stll dstrbuted n a small regon, and thus such a small area could be deemed as an obectve functon to be optmzed. For the case of two gven tracs, the correspondng ntersectons n the parameter space are plotted n Fg.3, and for the upper left expanded subfgure, whch corresponds to target 1, the mnmum and mum values of could be obtaned and then denoted by mn and, respectvely. Smlarly, the related mnmum and the mum values of are also found and denoted by mn and, respectvely ( m ) ( m ) ( mn, mn ) target ( rad ) target 3 S 1 1 (, ) (, ) 1 ( m ) target 1 target ( mn, mn ) S ( rad ) ( rad ) Fg. 3. A case of determnaton of obectve functon n the parameter space As a result, two rectangular blocs are formed and the area of each s calculated as and the obectve functon J s defned as S ( ) ( ), (1) mn mn J mn M r 1 S r( r1 r r3 r4 ) s. t r m r1 r r3 r4, m1 m m3 m4, (11) 1,..., 4; where r 1 r r 3 r 4 or m1 m m3 m4 s the possble trac n the trac space, M s the number of tracs to be ntalzed. Afterwards, the ants of dfferent tass wll be nvestgated, and t has the followng characterstcs: 1) The number of tass s equal to the one of tracs to be ntated, or equal to the one of targets of nterest. ) The tradtonal ACO algorthm bulds solutons n an ncremental way, but the proposed system of dfferent tass of ants bulds solutons n parallel way. Especally, n the proposed system of ants of dfferent tass, the thought of both collaboraton and competton between ants s consdered and ntroduced. For nstance, ants of the same tas search for foods n a collaboratve way, whle ants of dfferent tass wll compete wth each other durng establshng solutons.

9 Ant Colony Optmzaton 17 3) Ants of the same tas are dedcated to fndng ther best soluton, and a set of all best solutons found by ants of dfferent tass consttute the solutons to Eq. (11) we descrbe. 4) In the system of ants of dfferent tass, the search space depends not only on the measurement returns at the next scan but also on the pror nowledge of target moton. The determnaton of search space In the case of bearngs-only two-sensor- M -target tracng, the samplng data of the frst four scans are utlzed sequentally to ntate tracs, and then total four search spaces,.e., 1,, 3, and 4, are obtaned sequentally. Suppose that the pror nowledge about target moton, such as the mnmum and mum veloctes denoted by v respectvely, s nown and then utlzed to construct an annular regon whose nner and outer raduses are determned by r 1 v mn T and r v T, respectvely, where T denotes the samplng nterval. For nstance, f an ant s now located at poston n 1, then the ant wll vst the next poston located n the shadow secton covered by both the annular regon and, whch s denoted by n Fg.4. vmn and 1 r 1 r 3 4 Fg. 4. The determnaton of search spaces Trac Canddate Constructon Usng the Ants of Dfferent Tass Intally, M ants of dfferent tass are placed randomly on poston canddates n the frst search space 1, then each ant of a gven tas vsts probablstcally the poston canddate n the next search space. Suppose that an ant of a gven tas s s now located at poston n (1 3 ), then the ant wll vst poston n the next search space by applyng the followng probablstc formula:

10 18 New Advances n Machne Learnng s, 1 arg, f q q s 1,,, J otherwse, (1) and J s a random varable selected accordng to the followng probablty dstrbuton s, 1, s,,, f 1 P( ) s, l 1 (13) l, l s 1, l, l, l otherwse s where, denotes the pheromone amount deposted by ants of tas s on tral (, ),, s the total pheromone amount deposted by all ants of dfferent tass on tral (, ), shows the repulson on the foregn pheromones left on the tral (, ), q s a random number unformly dstrbuted between and 1, and q s a parameter whch determnes the relatve mportance of the explotaton of good solutons versus the exploraton of search spaces. Accordng to the search spaces dscussed above, Fg. 5 plots the process of how the heurstc value s calculated from search spaces 1 to, namely, f an ant wll move from postons to, the correspondng heurstc value can be defned as ( d r ), (14) r s equal to ( r r1 ) /.,, exp ( r r1 ) where d, denotes the dstance between postons and, and Note that f poston falls out of, we set,, and the search falure s declared for the current ant. r 1 r d,, r o r 1 r d, r r Fg. 5. The calculaton of heurstc value

11 Ant Colony Optmzaton 19 Update of Pheromone The pheromone update s performed n two phases, namely, local update and global update. Whle buldng a soluton, f an ant of tas s carres out the transton from postons to, then the pheromone level of the correspondng tral s changed n the followng way: (1 ), (15) s s s,, s where s the ntal pheromone level of ants of tas s. Once all ants of dfferent tass at a gven teraton have vsted four canddate postons each from dfferent samplng ndces, the pheromone amount on each establshed trac wll be updated globally. Here, we use the best-so-far-soluton found by ants of the same tas,.e. the best soluton found from the start of the algorthm run, to update the correspondng pheromone tral. We adopt the followng rule p s s s,, (1 ),, 1. (16) s, where s the pheromone amount that ant of tas s deposts on the tral (, ) t has, traveled at the current teraton, and p s the number of ants. In the case of bearngs-only s, mult-sensor-mult-target tracng, s set to a constant., 4. A Comparson of ACO-Based Methods for Trac Intaton 4.1 The Problem Two cases are nvestgated here, namely two and three tracs ntaton problems. For each scenaro, the performance of trac ntaton s nvestgated both n clutter-free envronments and n clutter envronments, respectvely. Two fxed sensors used to measure the targets bearngs are located at (, ) and (18 m, ) respectvely n a survellance regon. The standard devaton of the bearng measurements for each sensor s taen as.1, and the samplng nterval s set to be T 1s. The case n whch each target maes a unform rectlnear moton s consdered, and the ntal state of each target s llustrated n Table 1. Scenaros 1 Targets x y x y (m) (m) (m/s) (m/s) Table 1. The ntal poston and velocty of each target n the two consdered scenaros

12 New Advances n Machne Learnng 7 6 target 1 target ghost target 1 target target 3 ghost 5 45 Y (m) 4 Y (m) X (m) X (m) Fg. 6. The target poston canddates n a clutter-free envronment (left: Scenaro 1, rght: Scenaro ) target 1 target clutter and ghost 8 7 target 1 target target 3 clutter and ghost 45 6 Y (m) Y (m) X (m) X (m) Fg. 7. The target poston canddates n clutter envronments (left: Scenaro 1, rght: Scenaro ) Fgs.6 and 7 depct a part of poston canddates obtaned by ntersectng LOSs at each scan, and our obect s to dscrmnate the true postons of each target of nterest. Here, we use two ACO-based technques, namely the Ant System (called the tradtonal ACO) and the system of ants of dfferent tass (called the proposed ACO). Other parameters related to the two ACO-based methods are llustrated n Table Parameter Value Parameter Value.1.3. M 3M v mn 1 m / s.8 v 4 m / s q.7 a 15 m / s.5 N 5 Table. The Parameter Settngs for ACO-related Methods

13 Ant Colony Optmzaton 1 4. Evaluaton Indces Two performance ndces are ntroduced to evaluate the system of ants of dfferent tass,.e. The probablty of false trac ntaton: assumng N Monte-Carlo runs are performed, we defne the probablty of false trac ntaton as N N F f n, (17) 1 1 where f denotes the number of false ntated tracs at the th Monte-Carlo run, and n s the total number of ntated tracs. The probablty of correct ntaton of at least tracs: f at least (1 M ) tracs are ntated correctly, ts correspondng probablty s where l s a bnary varable and defned as N 1 C l N, (18) 1 l at the th Monte-Carlo run. f at least tracs are ntated correctly otherwse (19) 4.3 Results All results n Tables 3 to 6 are averaged over 1, Monte-Carlo runs. Accordng to the evaluaton ndces we ntroduce, the tradtonal ACO algorthm performs as well as the proposed one, as llustrated n Tables 3 and 4, n clutter-free envronments. However, n the presence of clutter, the proposed ACO algorthm shows a sgnfcant mprovement over the tradtonal one wth respect to the probablty of false trac ntaton, as shown n Tables 5 and 6. Evaluaton ndces The tradtonal ACO The proposed ACO Pro. of false trac ntaton ( F ).1. Pro. of correct ntaton of at least tracs( C ) C C Table 3. Performance comparson for two-trac-ntaton problem n clutter-free envronments

14 New Advances n Machne Learnng Evaluaton ndces The tradtonal ACO The proposed ACO Pro. of false trac ntaton ( F ) C Pro. of correct ntaton of at least tracs( C ) C C Table 4. Performance comparson for three-trac-ntaton problem n clutter-free envronments Evaluaton ndces The tradtonal ACO The proposed ACO Pro. of false trac ntaton ( F ) Pro. of correct ntaton of at least tracs( C ) C C Table 5. Performance comparson for two-trac-ntaton problem n clutter envronments Evaluaton ndces The tradtonal ACO The proposed ACO Pro. of false trac ntaton ( F ) C Pro. of correct ntaton of at least tracs( C ) C C Table 6. Performance comparson for three-trac-ntaton problem n clutter envronments Among 1, Monte-Carlo runs, only the cases of all tracs beng ntated successfully are nvestgated and called effectve runs later. For the obectvty of comparson, we select the worst case, n whch the mum runnng tme for each ACO algorthm s evaluated, from the effectve runs.

15 Ant Colony Optmzaton 3 Fg. 8 depcts the trends of obectve functon evoluton wth the ncreasng number of teratons n scenaro. Compared wth the tradtonal ACO algorthm, the proposed one requres fewer teratons for convergence n clutter-free or clutter envronments. Accordng to Tables 3 and 4, although the performance of the tradtonal ACO algorthm s comparable to that of the proposed one, we fnd that the proposed ACO one seems more practcal due to less runnng tme needed. Fgs. 9 and 1 depct varyng curves of pheromone on the true targets tracs, t s oerved that the amount of pheromone on each true trac ncreases n a moderate way, whch means most ants prefer choosng these tracs and regarded them as optmal solutons The proposed ACO The tradtonal ACO 35 The proposed ACO The tradtonal ACO 9 3 Obectve value (m.rad) Obectve value (m.rad) Iteraton Iteraton Fg. 8. Obectve functon curves (left: In clutter-free envronments; rght: In clutter envronments) Pheromone amount.75 On trac 1 On trac On trac 3 Pheromone amount On trac 1 On trac On trac Iteraton Iteraton Fg. 9. Pheromone curves n clutter-free envronments (left: The proposed ACO; rght: The tradtonal ACO)

16 4 New Advances n Machne Learnng On trac 1 On trac On trac Pheromone amount Iteraton Pheromone amount On trac 1 On trac On trac Iteraton Fg. 1. Pheromone curves n clutter envronments (left: The proposed ACO; rght: The tradtonal ACO) 5. Concluson Ths chapter manly ams to ntroduce some wdely used ACO algorthms and ther orgns, such as the AS, EAS, MMAS, and so on. It s found that all concerns are focused on the pheromone update strategy. Some uses the best-so-far-ant or the teraton-best ant ndependently/nteractvely to update the tral that ants travelled. Meanwhle, the update law may dffer a bt for dfferent ACO algorthms. Among the four ACO algorthms, two versons have receved great populartes n varous applcatons,.e. AS and MMAS. Another contrbuton n ths chapter s the extenson of the general ACO algorthm to the system of ants of dfferent tass, and ts behavour s modelled and mplemented n the trac ntaton problems. Smulaton results are also presented to show the effectveness of the novel ACO algorthm. Accordng to the example presented n ths chapter, we beleve that the general framewor of AS can be modfed to solve varous optmal or non-optmal problems. 6. References B. Bullnhemer; R. F. Hartl & C. Strauss. (1999). A new ran based verson of the ant system: A computatonal study, Central Eur. J. Oper. Res. Econ., Vol. 7, No. 1, 5 38, ISSN X. C. Blum. (5). Beam-ACO hybrdzng ant colony optmzaton wth beam search: An applcaton to open shop schedulng, Comput. Oper. Res., Vol. 3, No. 6, , ISSN Davd Martens; Manu De Bacer & Raf Haesen. (7). Classfcaton Wth Ant Colony Optmzaton, IEEE Trans. on Evolutonal Computaton, Vol. 11, No. 5, October 7, , ISSN X. Kutluyll Dogancay. (4). On the bas of lnear least squares algorthm for passve target localzaton, Sgnal Processng, Vol. 84, No. 3, , ISSN

17 Ant Colony Optmzaton 5 Kutluyll Dogancay. (5). Bearngs-only target localzaton usng total least squares, Sgnal Processng, Vol. 85, No. 9, , ISSN M. Dorgo; V. Manezzo & A. Colorn. (1991). Postve Feedbac as a Search Strategy, Techncal Report 91 16, Poltecnco d Mlano, Mlano, Italy. M. Dorgo; V. Manezzo & A. Colorn. (1996). The ant system: optmzaton by a colony of cooperatng agents, IEEE Trans. on System, Man, and Cybernetcs-part B, Vol.6, No. 1, 9-4, ISSN M. Dorgo & L. M. Gambardella. (1997). Ant colony system: A cooperatve learnng approach to the travelng salesman problem, IEEE Trans. on Evolutonal Computaton, Vol.1, No. 1, 53-66, ISSN X. P. Bhattacharya; A. Rosenfeld & I. Wess. (). Pont-to-lne mappngs as Hough transforms, Pattern Recognton Letters, Vol. 3, No. 4, , ISSN Parag M. Kanade & Lawrence O. Hall. (7). Fuzzy Ants and Clusterng, IEEE Trans. on System, Man, and Cybernetcs-part A, Vol. 37, No. 5, September 7, ,ISSN R. Montemann; L. M. Gambardella; A. E. Rzzol & A. Donat. (5). Ant colony system for a dynamc vehcle routng problem, J. Combnatoral Optm., Vol. 1, No. 4, , ISSN S.C. Nardone; A.G. Lndgren & K.F. Gong. (1984). Fundamental propertes and performance of conventonal bearngs-only target moton analyss, IEEE Transactons on Aerospace and Electronc Systems, Vol. 9, No. 9, , ISSN T. Stützle & H. H. Hoos. (). MAX-MIN ant system, Future generaton computer systems, Vol.16, , ISSN X. V. Shapro. (6). Accuracy of the straght lne Hough transform: the non-votng approach, Computer Vson and Image Understandng, Vol. 13, No. 1, 1-1, ISSN

18 6 New Advances n Machne Learnng

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

The Study of Teaching-learning-based Optimization Algorithm

The Study of Teaching-learning-based Optimization Algorithm Advanced Scence and Technology Letters Vol. (AST 06), pp.05- http://dx.do.org/0.57/astl.06. The Study of Teachng-learnng-based Optmzaton Algorthm u Sun, Yan fu, Lele Kong, Haolang Q,, Helongang Insttute

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Markov Chain Monte Carlo Lecture 6

Markov Chain Monte Carlo Lecture 6 where (x 1,..., x N ) X N, N s called the populaton sze, f(x) f (x) for at least one {1, 2,..., N}, and those dfferent from f(x) are called the tral dstrbutons n terms of mportance samplng. Dfferent ways

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

biologically-inspired computing lecture 21 Informatics luis rocha 2015 INDIANA UNIVERSITY biologically Inspired computing

biologically-inspired computing lecture 21 Informatics luis rocha 2015 INDIANA UNIVERSITY biologically Inspired computing lecture 21 -nspred Sectons I485/H400 course outlook Assgnments: 35% Students wll complete 4/5 assgnments based on algorthms presented n class Lab meets n I1 (West) 109 on Lab Wednesdays Lab 0 : January

More information

Third example: Ants foraging behavior as an inspiration for shortest path algorithms (ant colony optimization)

Third example: Ants foraging behavior as an inspiration for shortest path algorithms (ant colony optimization) From scentfc to engneerng swarm ntellgence Examples Thrd example: Ants foragng behavor as an nspraton for shortest path algorthms (ant colony optmzaton) Cemetery organzaton and brood sortng data clusterng

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

Outline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1]

Outline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1] DYNAMIC SHORTEST PATH SEARCH AND SYNCHRONIZED TASK SWITCHING Jay Wagenpfel, Adran Trachte 2 Outlne Shortest Communcaton Path Searchng Bellmann Ford algorthm Algorthm for dynamc case Modfcatons to our algorthm

More information

Chapter - 2. Distribution System Power Flow Analysis

Chapter - 2. Distribution System Power Flow Analysis Chapter - 2 Dstrbuton System Power Flow Analyss CHAPTER - 2 Radal Dstrbuton System Load Flow 2.1 Introducton Load flow s an mportant tool [66] for analyzng electrcal power system network performance. Load

More information

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS HCMC Unversty of Pedagogy Thong Nguyen Huu et al. A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS Thong Nguyen Huu and Hao Tran Van Department of mathematcs-nformaton,

More information

A linear imaging system with white additive Gaussian noise on the observed data is modeled as follows:

A linear imaging system with white additive Gaussian noise on the observed data is modeled as follows: Supplementary Note Mathematcal bacground A lnear magng system wth whte addtve Gaussan nose on the observed data s modeled as follows: X = R ϕ V + G, () where X R are the expermental, two-dmensonal proecton

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

Lecture 4. Instructor: Haipeng Luo

Lecture 4. Instructor: Haipeng Luo Lecture 4 Instructor: Hapeng Luo In the followng lectures, we focus on the expert problem and study more adaptve algorthms. Although Hedge s proven to be worst-case optmal, one may wonder how well t would

More information

Generalized Linear Methods

Generalized Linear Methods Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set

More information

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Design and Optimization of Fuzzy Controller for Inverse Pendulum System Using Genetic Algorithm

Design and Optimization of Fuzzy Controller for Inverse Pendulum System Using Genetic Algorithm Desgn and Optmzaton of Fuzzy Controller for Inverse Pendulum System Usng Genetc Algorthm H. Mehraban A. Ashoor Unversty of Tehran Unversty of Tehran h.mehraban@ece.ut.ac.r a.ashoor@ece.ut.ac.r Abstract:

More information

Chapter 9: Statistical Inference and the Relationship between Two Variables

Chapter 9: Statistical Inference and the Relationship between Two Variables Chapter 9: Statstcal Inference and the Relatonshp between Two Varables Key Words The Regresson Model The Sample Regresson Equaton The Pearson Correlaton Coeffcent Learnng Outcomes After studyng ths chapter,

More information

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle

More information

Annexes. EC.1. Cycle-base move illustration. EC.2. Problem Instances

Annexes. EC.1. Cycle-base move illustration. EC.2. Problem Instances ec Annexes Ths Annex frst llustrates a cycle-based move n the dynamc-block generaton tabu search. It then dsplays the characterstcs of the nstance sets, followed by detaled results of the parametercalbraton

More information

The Minimum Universal Cost Flow in an Infeasible Flow Network

The Minimum Universal Cost Flow in an Infeasible Flow Network Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

More information

Supporting Information

Supporting Information Supportng Informaton The neural network f n Eq. 1 s gven by: f x l = ReLU W atom x l + b atom, 2 where ReLU s the element-wse rectfed lnear unt, 21.e., ReLUx = max0, x, W atom R d d s the weght matrx to

More information

Uncertainty in measurements of power and energy on power networks

Uncertainty in measurements of power and energy on power networks Uncertanty n measurements of power and energy on power networks E. Manov, N. Kolev Department of Measurement and Instrumentaton, Techncal Unversty Sofa, bul. Klment Ohrdsk No8, bl., 000 Sofa, Bulgara Tel./fax:

More information

On the Multicriteria Integer Network Flow Problem

On the Multicriteria Integer Network Flow Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

Uncertainty and auto-correlation in. Measurement

Uncertainty and auto-correlation in. Measurement Uncertanty and auto-correlaton n arxv:1707.03276v2 [physcs.data-an] 30 Dec 2017 Measurement Markus Schebl Federal Offce of Metrology and Surveyng (BEV), 1160 Venna, Austra E-mal: markus.schebl@bev.gv.at

More information

Using Immune Genetic Algorithm to Optimize BP Neural Network and Its Application Peng-fei LIU1,Qun-tai SHEN1 and Jun ZHI2,*

Using Immune Genetic Algorithm to Optimize BP Neural Network and Its Application Peng-fei LIU1,Qun-tai SHEN1 and Jun ZHI2,* Advances n Computer Scence Research (ACRS), volume 54 Internatonal Conference on Computer Networks and Communcaton Technology (CNCT206) Usng Immune Genetc Algorthm to Optmze BP Neural Network and Its Applcaton

More information

CSC 411 / CSC D11 / CSC C11

CSC 411 / CSC D11 / CSC C11 18 Boostng s a general strategy for learnng classfers by combnng smpler ones. The dea of boostng s to take a weak classfer that s, any classfer that wll do at least slghtly better than chance and use t

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

A Bayes Algorithm for the Multitask Pattern Recognition Problem Direct Approach

A Bayes Algorithm for the Multitask Pattern Recognition Problem Direct Approach A Bayes Algorthm for the Multtask Pattern Recognton Problem Drect Approach Edward Puchala Wroclaw Unversty of Technology, Char of Systems and Computer etworks, Wybrzeze Wyspanskego 7, 50-370 Wroclaw, Poland

More information

AN IMPROVED PARTICLE FILTER ALGORITHM BASED ON NEURAL NETWORK FOR TARGET TRACKING

AN IMPROVED PARTICLE FILTER ALGORITHM BASED ON NEURAL NETWORK FOR TARGET TRACKING AN IMPROVED PARTICLE FILTER ALGORITHM BASED ON NEURAL NETWORK FOR TARGET TRACKING Qn Wen, Peng Qcong 40 Lab, Insttuton of Communcaton and Informaton Engneerng,Unversty of Electronc Scence and Technology

More information

Pop-Click Noise Detection Using Inter-Frame Correlation for Improved Portable Auditory Sensing

Pop-Click Noise Detection Using Inter-Frame Correlation for Improved Portable Auditory Sensing Advanced Scence and Technology Letters, pp.164-168 http://dx.do.org/10.14257/astl.2013 Pop-Clc Nose Detecton Usng Inter-Frame Correlaton for Improved Portable Audtory Sensng Dong Yun Lee, Kwang Myung Jeon,

More information

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests Smulated of the Cramér-von Mses Goodness-of-Ft Tests Steele, M., Chaselng, J. and 3 Hurst, C. School of Mathematcal and Physcal Scences, James Cook Unversty, Australan School of Envronmental Studes, Grffth

More information

Difference Equations

Difference Equations Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1

More information

Chapter Newton s Method

Chapter Newton s Method Chapter 9. Newton s Method After readng ths chapter, you should be able to:. Understand how Newton s method s dfferent from the Golden Secton Search method. Understand how Newton s method works 3. Solve

More information

Appendix B: Resampling Algorithms

Appendix B: Resampling Algorithms 407 Appendx B: Resamplng Algorthms A common problem of all partcle flters s the degeneracy of weghts, whch conssts of the unbounded ncrease of the varance of the mportance weghts ω [ ] of the partcles

More information

Speeding up Computation of Scalar Multiplication in Elliptic Curve Cryptosystem

Speeding up Computation of Scalar Multiplication in Elliptic Curve Cryptosystem H.K. Pathak et. al. / (IJCSE) Internatonal Journal on Computer Scence and Engneerng Speedng up Computaton of Scalar Multplcaton n Ellptc Curve Cryptosystem H. K. Pathak Manju Sangh S.o.S n Computer scence

More information

Tracking with Kalman Filter

Tracking with Kalman Filter Trackng wth Kalman Flter Scott T. Acton Vrgna Image and Vdeo Analyss (VIVA), Charles L. Brown Department of Electrcal and Computer Engneerng Department of Bomedcal Engneerng Unversty of Vrgna, Charlottesvlle,

More information

Online Classification: Perceptron and Winnow

Online Classification: Perceptron and Winnow E0 370 Statstcal Learnng Theory Lecture 18 Nov 8, 011 Onlne Classfcaton: Perceptron and Wnnow Lecturer: Shvan Agarwal Scrbe: Shvan Agarwal 1 Introducton In ths lecture we wll start to study the onlne learnng

More information

Credit Card Pricing and Impact of Adverse Selection

Credit Card Pricing and Impact of Adverse Selection Credt Card Prcng and Impact of Adverse Selecton Bo Huang and Lyn C. Thomas Unversty of Southampton Contents Background Aucton model of credt card solctaton - Errors n probablty of beng Good - Errors n

More information

arxiv:cs.cv/ Jun 2000

arxiv:cs.cv/ Jun 2000 Correlaton over Decomposed Sgnals: A Non-Lnear Approach to Fast and Effectve Sequences Comparson Lucano da Fontoura Costa arxv:cs.cv/0006040 28 Jun 2000 Cybernetc Vson Research Group IFSC Unversty of São

More information

A Particle Filter Algorithm based on Mixing of Prior probability density and UKF as Generate Importance Function

A Particle Filter Algorithm based on Mixing of Prior probability density and UKF as Generate Importance Function Advanced Scence and Technology Letters, pp.83-87 http://dx.do.org/10.14257/astl.2014.53.20 A Partcle Flter Algorthm based on Mxng of Pror probablty densty and UKF as Generate Importance Functon Lu Lu 1,1,

More information

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

More information

MMA and GCMMA two methods for nonlinear optimization

MMA and GCMMA two methods for nonlinear optimization MMA and GCMMA two methods for nonlnear optmzaton Krster Svanberg Optmzaton and Systems Theory, KTH, Stockholm, Sweden. krlle@math.kth.se Ths note descrbes the algorthms used n the author s 2007 mplementatons

More information

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for P Charts. Dr. Wayne A. Taylor

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for P Charts. Dr. Wayne A. Taylor Taylor Enterprses, Inc. Control Lmts for P Charts Copyrght 2017 by Taylor Enterprses, Inc., All Rghts Reserved. Control Lmts for P Charts Dr. Wayne A. Taylor Abstract: P charts are used for count data

More information

18.1 Introduction and Recap

18.1 Introduction and Recap CS787: Advanced Algorthms Scrbe: Pryananda Shenoy and Shjn Kong Lecturer: Shuch Chawla Topc: Streamng Algorthmscontnued) Date: 0/26/2007 We contnue talng about streamng algorthms n ths lecture, ncludng

More information

Portfolios with Trading Constraints and Payout Restrictions

Portfolios with Trading Constraints and Payout Restrictions Portfolos wth Tradng Constrants and Payout Restrctons John R. Brge Northwestern Unversty (ont wor wth Chrs Donohue Xaodong Xu and Gongyun Zhao) 1 General Problem (Very) long-term nvestor (eample: unversty

More information

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method Appled Mathematcal Scences, Vol. 7, 0, no. 47, 07-0 HIARI Ltd, www.m-hkar.com Comparson of the Populaton Varance Estmators of -Parameter Exponental Dstrbuton Based on Multple Crtera Decson Makng Method

More information

Interactive Bi-Level Multi-Objective Integer. Non-linear Programming Problem

Interactive Bi-Level Multi-Objective Integer. Non-linear Programming Problem Appled Mathematcal Scences Vol 5 0 no 65 3 33 Interactve B-Level Mult-Objectve Integer Non-lnear Programmng Problem O E Emam Department of Informaton Systems aculty of Computer Scence and nformaton Helwan

More information

DETERMINATION OF UNCERTAINTY ASSOCIATED WITH QUANTIZATION ERRORS USING THE BAYESIAN APPROACH

DETERMINATION OF UNCERTAINTY ASSOCIATED WITH QUANTIZATION ERRORS USING THE BAYESIAN APPROACH Proceedngs, XVII IMEKO World Congress, June 7, 3, Dubrovn, Croata Proceedngs, XVII IMEKO World Congress, June 7, 3, Dubrovn, Croata TC XVII IMEKO World Congress Metrology n the 3rd Mllennum June 7, 3,

More information

Lecture Space-Bounded Derandomization

Lecture Space-Bounded Derandomization Notes on Complexty Theory Last updated: October, 2008 Jonathan Katz Lecture Space-Bounded Derandomzaton 1 Space-Bounded Derandomzaton We now dscuss derandomzaton of space-bounded algorthms. Here non-trval

More information

Global Sensitivity. Tuesday 20 th February, 2018

Global Sensitivity. Tuesday 20 th February, 2018 Global Senstvty Tuesday 2 th February, 28 ) Local Senstvty Most senstvty analyses [] are based on local estmates of senstvty, typcally by expandng the response n a Taylor seres about some specfc values

More information

Some modelling aspects for the Matlab implementation of MMA

Some modelling aspects for the Matlab implementation of MMA Some modellng aspects for the Matlab mplementaton of MMA Krster Svanberg krlle@math.kth.se Optmzaton and Systems Theory Department of Mathematcs KTH, SE 10044 Stockholm September 2004 1. Consdered optmzaton

More information

CHAPTER 14 GENERAL PERTURBATION THEORY

CHAPTER 14 GENERAL PERTURBATION THEORY CHAPTER 4 GENERAL PERTURBATION THEORY 4 Introducton A partcle n orbt around a pont mass or a sphercally symmetrc mass dstrbuton s movng n a gravtatonal potental of the form GM / r In ths potental t moves

More information

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced,

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced, FREQUENCY DISTRIBUTIONS Page 1 of 6 I. Introducton 1. The dea of a frequency dstrbuton for sets of observatons wll be ntroduced, together wth some of the mechancs for constructng dstrbutons of data. Then

More information

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for U Charts. Dr. Wayne A. Taylor

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for U Charts. Dr. Wayne A. Taylor Taylor Enterprses, Inc. Adjusted Control Lmts for U Charts Copyrght 207 by Taylor Enterprses, Inc., All Rghts Reserved. Adjusted Control Lmts for U Charts Dr. Wayne A. Taylor Abstract: U charts are used

More information

Boostrapaggregating (Bagging)

Boostrapaggregating (Bagging) Boostrapaggregatng (Baggng) An ensemble meta-algorthm desgned to mprove the stablty and accuracy of machne learnng algorthms Can be used n both regresson and classfcaton Reduces varance and helps to avod

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

A METHOD FOR DETECTING OUTLIERS IN FUZZY REGRESSION

A METHOD FOR DETECTING OUTLIERS IN FUZZY REGRESSION OPERATIONS RESEARCH AND DECISIONS No. 2 21 Barbara GŁADYSZ* A METHOD FOR DETECTING OUTLIERS IN FUZZY REGRESSION In ths artcle we propose a method for dentfyng outlers n fuzzy regresson. Outlers n a sample

More information

Average Decision Threshold of CA CFAR and excision CFAR Detectors in the Presence of Strong Pulse Jamming 1

Average Decision Threshold of CA CFAR and excision CFAR Detectors in the Presence of Strong Pulse Jamming 1 Average Decson hreshold of CA CFAR and excson CFAR Detectors n the Presence of Strong Pulse Jammng Ivan G. Garvanov and Chrsto A. Kabachev Insttute of Informaton echnologes Bulgaran Academy of Scences

More information

Inductance Calculation for Conductors of Arbitrary Shape

Inductance Calculation for Conductors of Arbitrary Shape CRYO/02/028 Aprl 5, 2002 Inductance Calculaton for Conductors of Arbtrary Shape L. Bottura Dstrbuton: Internal Summary In ths note we descrbe a method for the numercal calculaton of nductances among conductors

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement Markov Chan Monte Carlo MCMC, Gbbs Samplng, Metropols Algorthms, and Smulated Annealng 2001 Bonformatcs Course Supplement SNU Bontellgence Lab http://bsnuackr/ Outlne! Markov Chan Monte Carlo MCMC! Metropols-Hastngs

More information

Building A Fuzzy Inference System By An Extended Rule Based Q-Learning

Building A Fuzzy Inference System By An Extended Rule Based Q-Learning Buldng A Fuzzy Inference System By An Extended Rule Based Q-Learnng Mn-Soeng Km, Sun-G Hong and Ju-Jang Lee * Dept. of Electrcal Engneerng and Computer Scence, KAIST 373- Kusung-Dong Yusong-Ku Taejon 35-7,

More information

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES BÂRZĂ, Slvu Faculty of Mathematcs-Informatcs Spru Haret Unversty barza_slvu@yahoo.com Abstract Ths paper wants to contnue

More information

Discrete Particle Swarm Optimization for TSP: Theoretical Results and Experimental Evaluations

Discrete Particle Swarm Optimization for TSP: Theoretical Results and Experimental Evaluations Dscrete Partcle Swarm Optmzaton for TSP: Theoretcal Results and Expermental Evaluatons Matthas Hoffmann, Mortz Mühlenthaler, Sabne Helwg, Rolf Wanka Department of Computer Scence, Unversty of Erlangen-Nuremberg,

More information

Société de Calcul Mathématique SA

Société de Calcul Mathématique SA Socété de Calcul Mathématque SA Outls d'ade à la décson Tools for decson help Probablstc Studes: Normalzng the Hstograms Bernard Beauzamy December, 202 I. General constructon of the hstogram Any probablstc

More information

A Hybrid Variational Iteration Method for Blasius Equation

A Hybrid Variational Iteration Method for Blasius Equation Avalable at http://pvamu.edu/aam Appl. Appl. Math. ISSN: 1932-9466 Vol. 10, Issue 1 (June 2015), pp. 223-229 Applcatons and Appled Mathematcs: An Internatonal Journal (AAM) A Hybrd Varatonal Iteraton Method

More information

Complement of Type-2 Fuzzy Shortest Path Using Possibility Measure

Complement of Type-2 Fuzzy Shortest Path Using Possibility Measure Intern. J. Fuzzy Mathematcal rchve Vol. 5, No., 04, 9-7 ISSN: 30 34 (P, 30 350 (onlne Publshed on 5 November 04 www.researchmathsc.org Internatonal Journal of Complement of Type- Fuzzy Shortest Path Usng

More information

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin Proceedngs of the 007 Wnter Smulaton Conference S G Henderson, B Bller, M-H Hseh, J Shortle, J D Tew, and R R Barton, eds LOW BIAS INTEGRATED PATH ESTIMATORS James M Calvn Department of Computer Scence

More information

SOLVING CAPACITATED VEHICLE ROUTING PROBLEMS WITH TIME WINDOWS BY GOAL PROGRAMMING APPROACH

SOLVING CAPACITATED VEHICLE ROUTING PROBLEMS WITH TIME WINDOWS BY GOAL PROGRAMMING APPROACH Proceedngs of IICMA 2013 Research Topc, pp. xx-xx. SOLVIG CAPACITATED VEHICLE ROUTIG PROBLEMS WITH TIME WIDOWS BY GOAL PROGRAMMIG APPROACH ATMII DHORURI 1, EMIUGROHO RATA SARI 2, AD DWI LESTARI 3 1Department

More information

CHAPTER IV RESEARCH FINDING AND DISCUSSIONS

CHAPTER IV RESEARCH FINDING AND DISCUSSIONS CHAPTER IV RESEARCH FINDING AND DISCUSSIONS A. Descrpton of Research Fndng. The Implementaton of Learnng Havng ganed the whole needed data, the researcher then dd analyss whch refers to the statstcal data

More information

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming EEL 6266 Power System Operaton and Control Chapter 3 Economc Dspatch Usng Dynamc Programmng Pecewse Lnear Cost Functons Common practce many utltes prefer to represent ther generator cost functons as sngle-

More information

Notes on Frequency Estimation in Data Streams

Notes on Frequency Estimation in Data Streams Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Adiabatic Sorption of Ammonia-Water System and Depicting in p-t-x Diagram

Adiabatic Sorption of Ammonia-Water System and Depicting in p-t-x Diagram Adabatc Sorpton of Ammona-Water System and Depctng n p-t-x Dagram J. POSPISIL, Z. SKALA Faculty of Mechancal Engneerng Brno Unversty of Technology Techncka 2, Brno 61669 CZECH REPUBLIC Abstract: - Absorpton

More information

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family IOSR Journal of Mathematcs IOSR-JM) ISSN: 2278-5728. Volume 3, Issue 3 Sep-Oct. 202), PP 44-48 www.osrjournals.org Usng T.O.M to Estmate Parameter of dstrbutons that have not Sngle Exponental Famly Jubran

More information

CIS526: Machine Learning Lecture 3 (Sept 16, 2003) Linear Regression. Preparation help: Xiaoying Huang. x 1 θ 1 output... θ M x M

CIS526: Machine Learning Lecture 3 (Sept 16, 2003) Linear Regression. Preparation help: Xiaoying Huang. x 1 θ 1 output... θ M x M CIS56: achne Learnng Lecture 3 (Sept 6, 003) Preparaton help: Xaoyng Huang Lnear Regresson Lnear regresson can be represented by a functonal form: f(; θ) = θ 0 0 +θ + + θ = θ = 0 ote: 0 s a dummy attrbute

More information

The equation of motion of a dynamical system is given by a set of differential equations. That is (1)

The equation of motion of a dynamical system is given by a set of differential equations. That is (1) Dynamcal Systems Many engneerng and natural systems are dynamcal systems. For example a pendulum s a dynamcal system. State l The state of the dynamcal system specfes t condtons. For a pendulum n the absence

More information

An Improved multiple fractal algorithm

An Improved multiple fractal algorithm Advanced Scence and Technology Letters Vol.31 (MulGraB 213), pp.184-188 http://dx.do.org/1.1427/astl.213.31.41 An Improved multple fractal algorthm Yun Ln, Xaochu Xu, Jnfeng Pang College of Informaton

More information

x i1 =1 for all i (the constant ).

x i1 =1 for all i (the constant ). Chapter 5 The Multple Regresson Model Consder an economc model where the dependent varable s a functon of K explanatory varables. The economc model has the form: y = f ( x,x,..., ) xk Approxmate ths by

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

An Application of Fuzzy Hypotheses Testing in Radar Detection

An Application of Fuzzy Hypotheses Testing in Radar Detection Proceedngs of the th WSES Internatonal Conference on FUZZY SYSEMS n pplcaton of Fuy Hypotheses estng n Radar Detecton.K.ELSHERIF, F.M.BBDY, G.M.BDELHMID Department of Mathematcs Mltary echncal Collage

More information

4DVAR, according to the name, is a four-dimensional variational method.

4DVAR, according to the name, is a four-dimensional variational method. 4D-Varatonal Data Assmlaton (4D-Var) 4DVAR, accordng to the name, s a four-dmensonal varatonal method. 4D-Var s actually a drect generalzaton of 3D-Var to handle observatons that are dstrbuted n tme. The

More information

Chapter 13: Multiple Regression

Chapter 13: Multiple Regression Chapter 13: Multple Regresson 13.1 Developng the multple-regresson Model The general model can be descrbed as: It smplfes for two ndependent varables: The sample ft parameter b 0, b 1, and b are used to

More information

Calculation of time complexity (3%)

Calculation of time complexity (3%) Problem 1. (30%) Calculaton of tme complexty (3%) Gven n ctes, usng exhaust search to see every result takes O(n!). Calculaton of tme needed to solve the problem (2%) 40 ctes:40! dfferent tours 40 add

More information

Multigradient for Neural Networks for Equalizers 1

Multigradient for Neural Networks for Equalizers 1 Multgradent for Neural Netorks for Equalzers 1 Chulhee ee, Jnook Go and Heeyoung Km Department of Electrcal and Electronc Engneerng Yonse Unversty 134 Shnchon-Dong, Seodaemun-Ku, Seoul 1-749, Korea ABSTRACT

More information

Neural networks. Nuno Vasconcelos ECE Department, UCSD

Neural networks. Nuno Vasconcelos ECE Department, UCSD Neural networs Nuno Vasconcelos ECE Department, UCSD Classfcaton a classfcaton problem has two types of varables e.g. X - vector of observatons (features) n the world Y - state (class) of the world x X

More information

Week 5: Neural Networks

Week 5: Neural Networks Week 5: Neural Networks Instructor: Sergey Levne Neural Networks Summary In the prevous lecture, we saw how we can construct neural networks by extendng logstc regresson. Neural networks consst of multple

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

An Interactive Optimisation Tool for Allocation Problems

An Interactive Optimisation Tool for Allocation Problems An Interactve Optmsaton ool for Allocaton Problems Fredr Bonäs, Joam Westerlund and apo Westerlund Process Desgn Laboratory, Faculty of echnology, Åbo Aadem Unversty, uru 20500, Fnland hs paper presents

More information

COS 521: Advanced Algorithms Game Theory and Linear Programming

COS 521: Advanced Algorithms Game Theory and Linear Programming COS 521: Advanced Algorthms Game Theory and Lnear Programmng Moses Charkar February 27, 2013 In these notes, we ntroduce some basc concepts n game theory and lnear programmng (LP). We show a connecton

More information

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling Real-Tme Systems Multprocessor schedulng Specfcaton Implementaton Verfcaton Multprocessor schedulng -- -- Global schedulng How are tasks assgned to processors? Statc assgnment The processor(s) used for

More information