SlideShare a Scribd company logo
1 of 41
Download to read offline
CIG case study: car racing
 ā€¢ A prolonged example of applying CI to a
   game: car racing
 ā€¢ Sensor representation and input selection
 ā€¢ Incremental evolution
 ā€¢ Competitive coevolution
 ā€¢ Player modelling
 ā€¢ Content creation
Racing games
ā€¢ On the charts for the last three decades
ā€¢ Can be technically simple (computationally
  cheap) or very sophisticated
ā€¢ Easy to pick up and play, but possess almost
  unlimited ā€œdepthā€ (a lifetime to master)
ā€¢ Can be played on your own or with others
CI in racing games
ā€¢ Learning to race
 ā€¢ on your own, against speciļ¬c opponents,
    against opponents in general, on one or
    several tracks, using simple or complex
    cars/physics models, etc.
ā€¢ Modelling driving styles
ā€¢ Creating entertaining game content: tracks
  and opponent drivers
A simple car game


ā€¢ Optimised for speed, not for prettiness
ā€¢ 2D dynamics (momentum, understeer, etc.)
ā€¢ Intended to qualitatively replicate a
  standard toy R/C car driven on a table
ā€¢ Bang-bang control (9 possible commands)
ā€¢ Walls are solid
ā€¢ Waypoints must be
  passed in order

ā€¢ Fitness: continuous
  approximation of
  waypoints passed in
  700 time steps
ā€¢ Inputs
 ā€¢ Six range-ļ¬nder sensors (evolvable pos.)
 ā€¢ Waypoint sensor, Speed, Bias
ā€¢ Networks
 ā€¢ Standard MLP, 9:6:2
 ā€¢ Outputs interpreted as thrust/steering
T rack        10                 50                  100                 200               P r.
                                           1             1.9 (0.1)          1.99 (0.06)         2.02    (0.01)      2.04 (0.02)       10
                                           2             2.06 (0.1)         2.12 (0.04)         2.14    (0)         2.15 (0.01)       10
                                           3             3.25 (0.08)        3.4 (0.1)           3.45    (0.12)      3.57 (0.1)        10
                                           4             3.35 (0.11)        3.58 (0.11)         3.61    (0.1)       3.67 (0.1)        10
                                           5             2.66 (0.13)        2.84 (0.02)         2.88    (0.06)      2.88 (0.06)       10
                                           6             2.64 (0)           2.71 (0.08)         2.72    (0.08)      2.82 (0.1)        10
                                           7             1.53 (0.29)        1.84 (0.13)         1.88    (0.12)      1.9 (0.09)        10
                       T rack             18                2
                                                         0.59 (0.15)           3
                                                                            0.73 (0.22)            4
                                                                                                0.85    (0.21)        5
                                                                                                                    0.93 (0.25)       06                 7               8
                       Fitness (sd)       1.66 (0.12)        1.86 (0.02)         2.27 (0.45) 2.66 (0.3)
                                                                                   TABLE VI                           2.19 (0.23)        2.47 (0.18)     0.22 (0.15)     0.15 (0.01)
                                                                               TABLE V
                                             F ITNESS OF BEST CONTROLLERS , EVOLVING CONTROLLERS
                     F ITNESS OF A FURTHER EVOLVED GENERAL CONTROLLER WITH EVOLVABLE SENSOR PARAMETERS ON THE DIFFERENT TRACKS . C OMPOUND FITNESS
                                       SPECIALISED FOR EACH TRACK , STARTING FROM A FURTHER EVOLVED
                                                                                                    2.22 (0.09).
                                                GENERAL CONTROLLER WITH EVOLVED SENSOR PARAMETERS .
                                                                                                                                                 Fig. 6. Sensor setup of a controller
                                                                                                                                                 reach good ļ¬tness on, track 7. Presum
                                                                                                                                                 their angular spread reļ¬‚ects the larg
                       T rack     10              50             100              200               P r.
                                                                            T rack    10              50            100           200           Phas to handle in order to navigate th
                                                                                                                                                  r.
                       1          1.9 (0.1)       1.99 (0.06)    2.02   (0.01)
                                                                            1          (0.02)
                                                                                  2.040.32 (0.07)   100.54 (0.2)    0.7 (0.38)    0.81 (0.5)    2
                       2          2.06 (0.1)      2.12 (0.04)    2.14   (0) 2     2.150.38 (0.24)
                                                                                       (0.01)       100.49 (0.38)   0.56 (0.36)   0.71 (0.5)    2
                       3          3.25 (0.08)     3.4 (0.1)      3.45   (0.12)
                                                                            3          (0.1)
                                                                                  3.570.32 (0.09)   100.97 (0.5)    1.47 (0.63)   1.98 (0.66)   7
                       4          3.35 (0.11)     3.58 (0.11)    3.61   (0.1)
                                                                            4          (0.1)
                                                                                  3.670.53 (0.17)   101.3 (0.48)    1.5 (0.54)    2.33 (0.59)   9
                       5          2.66 (0.13)     2.84 (0.02)    2.88       5
                                                                        (0.06)    2.880.45 (0.08)
                                                                                       (0.06)       100.95 (0.6)    0.95 (0.58)   1.65 (0.45)   8
                       6          2.64 (0)        2.71 (0.08)    2.72       6
                                                                        (0.08)        0.4 (0.08)
                                                                                  2.82 (0.1)        100.68 (0.27)   1.02 (0.74)   1.29 (0.76)   5
                       7          1.53 (0.29)     1.84 (0.13)    1.88       7
                                                                        (0.12)    1.9 0.3 (0.07)
                                                                                      (0.09)        100.35 (0.05)   0.39 (0.09)   0.46 (0.13)   0
                                                                            8         0.16 (0.02)     0.19 (0.03)   0.2 (0.01)    0.2 (0.01)    0
                       8          0.59 (0.15)     0.73 (0.22)    0.85   (0.21)    0.93 (0.25)       0
                                                                                                           TABLE I
                                                       TABLE VI
                                                                T HE FITNESS OF THE BEST CONTROLLER OF VARIOUS GENERATIONS ON
                           F ITNESS OF BEST CONTROLLERS , EVOLVING CONTROLLERS TRACKS , AND NUMBER OF RUNS PRODUCING
                                                                     THE DIFFERENT
                      SPECIALISED FOR EACH TRACK , STARTING FROMPROFICIENT CONTROLLERS . F ITNESS AVERAGED OVER 10 SEPARATE
                                                                   A FURTHER EVOLVED
                               GENERAL CONTROLLER WITH EVOLVED SENSOR PARAMETERS . STANDARD DEVIATION BETWEEN PARENTHESES .
                                                                                  EVOLUTIONARY RUNS ;
Fig. 2. The initial sensor setup, which is kept throughout the evolutionary                                  Fig. 6. 5.
                                                                                                              track Sensor setup of a controller specialized for, and able to consistently
run for those runs where sensor parameters are not evolvable. Here, setup of controller specialized forreach good While on, track 7. Presumably the use of all but one sensor and
                                                 Fig. 5. Sensor the car                                                         more or
                                                 less retaining the two longest-range sensors from the further evolved   ļ¬tness general
is seen in close-up moving upward-leftward. At this particular position, the
                                                                                                             their angular spread reļ¬‚ects the large variety of different situations the car
front-right sensor returns a positive number very close to 0, as it detects on, it has added medium-range sensors in the front and
                                                 controller it is based a                                    has to handle in order to navigate this more difļ¬cult track.
wall near the limit of its range; the front-left sensor returns a number close
                                                 back,The front, very short-range sensor to the left. number of waypoints in the track, 7. Sensor setup of another con
to 0.5, and the back sensor a slightly larger number.    and a left and right passed, divided by the                                            Fig.
sensors do not detect any walls at all and thus return 0.                        plus an intermediate term representing how far it is on its way in ļ¬gure 6 seemingly using all i
                                                                                                                                                one
                                                             to the next waypoint, calculated from the relative distances
                                                             between the car and the previous and next waypoint. A
range 200 pixels, as has three sensors pointing forward- ļ¬tness of 10 evolutionary runs were made, track
                                     controllers. For each track, 1.0 thus means having completed one full
                                                                                                                        VII. O BSERVATIONS ON EV
left, forward-right and backward respectively. The two other within the alloted time. Waypoints can only be passed in the
                                     where the initial population was seeded with the general
sensors, which point left and right, have reach 100; this is correct order, and a waypoint is counted as passed when the
illustrated in ļ¬gure 2.              controller and evolution of the car is within 30continue for waypoint. In It has previously been found
                                                             centre was allowed to pixels from the 200
Example video




Evolved with 50+50 ES, 100 Generatons
Choose your inputs
(+their representation)
ā€¢ Using third-person inputs (cartesian inputs)
  seems not to work
ā€¢ Either range-ļ¬nders or waypoint sensor
  can be taken away, but some ļ¬tness lost
ā€¢ A little bit of noise is not a problem,
  actually itā€™s desirable
ā€¢ Adding extra inputs (while keeping core
  inputs) can reduce evolvability drastically!
If you donā€™t know
       your inputs...
ā€¢ Memetic techniques (e.g. memetic ES) can
  sort out useful from useless inputs
ā€¢ Principle: evolve neural network weights
  together with a mask: whether connections
  are on or off
  ā€¢ Masks and weights are evolved at
    different time scales; after every mask
    mutation, weight space is searched - if no
    ļ¬tness increase, the mask is reverted
Learning controllers with irrelevant inputs present




   Togelius, Gomez and Schmidhuber (2008)
Generalization and
     specialization
ā€¢ A controller evolved for one track does
  not necessarily perform well on other
  tracks
ā€¢ How do we achieve more general game-
  playing skills?
 ā€¢ Is there a tradeoff between generality and
    performance?
damaging such cars in collisions is ha
                                                                            weight.
                                                                               The dynamics of the car are based on
                                                                            mechanical model, taking into account
                                                                            car and bad grip on the surface, but is n
                                                                            measurement [13][14]. The model is s
                                                                            [4], and differs mainly in its improve
                                                                            after more experience with the physical
                                                                            response system was reimplemented to
                                                                            realistic (and, as an effect, more undesir
                                                                            may cause the car to get stuck if the
                                                                            unfortunate angle, something often see
                                                                            physical cars.
                                                                               A track consists of a set of walls, a
                                                                            and a set of starting positions and di
                                                                            is added to a track in one of the sta
                                                                            corresponding starting direction, both t
                                                                            being subject to random alterations. Th
                                                                            for ļ¬tness calculations.
                                                                               For the experiments we have des
                                                                            tracks, presented in ļ¬gure 1. The tr
                                                                            vary in difļ¬culty, from easy to hard.
                                                                            are versions of three other tracks wi
                                                                            in reverse order, and the directions of
                                                                            reversed.
                                                                               The main differences between our
                                                                            real R/C car racing problem have to
                                                                            reported in Tanev et al. as well as [4]
                                                                            not unimportant lag in the communica
                                                                            computer and car, leading to the control
                                                                            perceptions. Apart from that, there
Fig. 1. The eight tracks. Notice how tracks 1 and 2 (at the top), 3 and
4, 5 and 6 differ in the clockwise/anti-clockwise layout of waypoints and   in estimations of the carā€™s position a
associated starting points. Tracks 7 and 8 have no relation to each other   overhead camera. In contrast, the sim
Incremental evolution
ā€¢ Introduced by Gomez & Mikkulainen (1997)
ā€¢ Change the ļ¬tness function f (to make it
  more demanding) as soon as a certain
  ļ¬tness is achieved
ā€¢ In this case, add new tracks to f as soon as
  the controller can drive 1.5 rounds on all
  tracks currently in f
Incremental evolution
ā€¢ Controllers evolved for speciļ¬c tracks
  perform poorly on other tracks
ā€¢ General controllers, that can drive almost
  any track, can be incrementally evolved
ā€¢ Starting from a general controller, a
  controller can be further evolved for
  specialization on a particular track
  ā€¢ drive faster than the general controller
  ā€¢ works even when evolution from scratch
    did not work!
Two cars on a track
ā€¢ Two car with solo-evolved controllers on
  one track: disaster
  ā€¢ they donā€™t even see each other!
ā€¢ How do we train controllers that take
  other drivers into account? (avoiding
  collisions or using them to their advantage)
ā€¢ Solution: car sensors (rangeļ¬nders, like the
  wall sensors) and competitive coevolution
Video: navigating
a complex track
Competitive
          coevolution
ā€¢ The ļ¬tness function evaluates at least two
  individuals
ā€¢ One individualā€™s success is adversely affected
  by the otherā€™s (directly or indirectly)
ā€¢ Very potent, but seldom straightforward;
  e.g. Hillis (1991), Rosin and Belew (1996)
Competitive
          coevolution
ā€¢ Standard 15+15 ES; each individual is
  evaluated through testing against the
  current best individual in the population
ā€¢ Fitness function a mix of...
 ā€¢ Absolute ļ¬tness: progress in n time steps
 ā€¢ Relative ļ¬tness: distance ahead of or
    behind the other car after n time steps
Video: absolute ļ¬tness
Video: 50/50 ļ¬tness
Video: relative ļ¬tness
Problems with
        coevolution
ā€¢ Over-specialization and cycling
 ā€¢ Can be battled with e.g. archives
ā€¢ Loss of gradient
 ā€¢ Can be battled with careful ļ¬tness
    function design, e.g. combining absolute
    and relative ļ¬tness
ā€¢ Much more research needed here!
Multi-population
      coevolution
ā€¢ Typically, competitive coevolution uses one
  or two populations
ā€¢ Many more populations can be used!
 ā€¢ Can help against cycling and
    overspecialization
  ā€¢ The phenotypical diversity between
    populations can be useful in itself
Example:
1 versus 9 populations




   Togelius, Burrow, Lucas (2007)
Player modelling
ā€¢ Can we create players that drive just like
  speciļ¬c human players?
ā€¢ The models need to be...
 ā€¢ Similar in terms of performance
 ā€¢ Similar in terms playing (driving) style
 ā€¢ Robust
Direct modelling
ā€¢ Let a player drive a number of tracks
ā€¢ Use supervised learning to associate inputs
  (sensors) with outputs (driving commands)
  ā€¢ e.g. MLP/Backpropagation or k-nearest
    neighbour
ā€¢ Suffers from generalization problems, and
  that any approximation is likely to lead to
  worse playing performance
Indirect modelling
ā€¢ Let a human drive a test track, record
  performance, speed and orthogonal
  deviation at the various waypoints the track
ā€¢ Start from a good, general evolved neural
  network controller, and evolve it further
ā€¢ Fitness: negative difference between
  controller and player for the three
  measures above
The test track supposedly requires
a varied repertoire of driving skills
                                                                                               1

                                                                                             0.8

                                                                                             0.6




                                                                 Fitness (progess, speed)
                                                                                             0.4

                                                                                             0.2

                                                                                               0

                                                                                            āˆ’0.2

                                                                                            āˆ’0.4

                                                                                            āˆ’0.6

                                                                                            āˆ’0.8
                                                                                                   0
                                                                                                   0        10


                Fig. 2.   The test track and the car.
                                                                                            Fig. 3.    Evolving a

   First of all, we design a test track, featuring a number of
different types of racing challenges. The track, as pictured                                  0

in (ļ¬g), has two long straight sections where the player can                          āˆ’0.2
Content creation
ā€¢ Creating interesting, enjoyable levels, worlds,
  tracks, opponents etc.
  ā€¢ Not the same as well-playing opponents
ā€¢ Probably the area where commercial game
  developers need most help
ā€¢ What makes game content fun? Many
  theories, e.g. Thomas Malone, Raph Koster,
  MihƔly CsƭkszentmihƔlyi
Track evolution
ā€¢ Using the controllers we evolved to model
  human players, we evolve tracks that are fun
  to drive for the modelled player
ā€¢ Fitness function:
 ā€¢ Right amount of progress
 ā€¢ Variation in progress
 ā€¢ High maximum speed
Fig. 5.   Track evolved using the random walk initialisation and mutation.
 e the representa-
nted with several
 t the beginning
plementations of
 nļ¬gurations are

rd initial track
 rners. Each mu-
 ontrol points by
distribution with
y axes.
 xperiments, mu-
 onļ¬guration, but
ectangle track is
 eds of mutations
  those mutations
 controller is not
e result of such a
 ll drivable track.
ck and evolution      Fig. 6. A track evolved (using the radial method) to be fun for the ļ¬rst
                      author, who plays too many racing games anyway. It is not easy to drive,
                      which is just as it should be.
n, starts from an
rol points around
the results of ou
                                                                              car racing [10].
                                                                                 In the section
                                                                              describe a numb
                                                                              value, most of wh
                                                                              described here. D
                                                                              sures would deļ¬n
                                                                              urgent to study th
                                                                              oft-cited hypothe
                                                                              know there are n
                                                                              entertainment me
                                                                              games and types
                                                                              needed.
                                                                                 Finally we no
                                                                              different approach
                                                                              in the beginning
Fig. 7. A track evolved (using the radial method) to be fun for the second
                                                                              viewed from sev
author, who is a bit more careful in his driving. Note the absence of sharp   on using evoluti
turns.                                                                        in games is not
                                                                              studying under w
                                                                              perspective we h
ks by sampling
aken advantage
 ack. First thick
  side of the b-
ixels or subject
nt is set up. But
 th of the track,
  and sometimes
struction of the
ing the b-spline
   middle of the
 imately regular
e resulting track
est track which
 e control points



                    Fig. 5.   Track evolved using the random walk initialisation and mutation.
 the representa-
ed with several
  the beginning
but only sometimes causes the car to collide. Those elements are believed to
be the main source of ļ¬nal progress variability. These features are also notably
absent from track c, on which the good player model has very low variability.
The progress of the controller is instead limited by many broad curves.




Fig. 3. Three evolved tracks: ((a)) evolved for a bad player with target progress 1.1,
(b) evolved for a good player with target ļ¬tness 1.5, (c) evolved for a good player with
target progress 1.5 using only progress ļ¬tness.
Video: evolved
TORCS drivers
Video: real car control
More on these topics
ā€¢ http://julian.togelius.com
 ā€¢ e.g. Togelius, Lucas and De Nardi:
    ā€œComputational Intelligence in Racing Gamesā€
ā€¢ Togelius, Gomez and Schmidhuber: ā€œLearning
  what to ignoreā€ on Friday, 11.10, room 606
ā€¢ Car Racing Competition on Tuesday 15.00,
  room 402

More Related Content

What's hot

Transmision manual hiunday
Transmision manual  hiundayTransmision manual  hiunday
Transmision manual hiundayDavid Parari
Ā 
Angry birds presentation
Angry birds presentationAngry birds presentation
Angry birds presentationlinhvu28
Ā 
Bolt details
Bolt detailsBolt details
Bolt detailsRamesh Naik
Ā 
Data processing Lab Lecture
Data processing Lab LectureData processing Lab Lecture
Data processing Lab Lecturewaddling
Ā 
Dolphin System Selling Points
Dolphin System Selling PointsDolphin System Selling Points
Dolphin System Selling Pointsguest17cd6
Ā 
punya iroh suniroh Metode Pengajaran Paket C Data Anova
punya iroh suniroh Metode Pengajaran Paket C Data Anovapunya iroh suniroh Metode Pengajaran Paket C Data Anova
punya iroh suniroh Metode Pengajaran Paket C Data Anovaguest5b160ded
Ā 
Clock Jitter and Measurement
Clock Jitter and MeasurementClock Jitter and Measurement
Clock Jitter and MeasurementSiTime Corporation
Ā 

What's hot (9)

Transmision manual hiunday
Transmision manual  hiundayTransmision manual  hiunday
Transmision manual hiunday
Ā 
Angry birds presentation
Angry birds presentationAngry birds presentation
Angry birds presentation
Ā 
Bolt details
Bolt detailsBolt details
Bolt details
Ā 
Data processing Lab Lecture
Data processing Lab LectureData processing Lab Lecture
Data processing Lab Lecture
Ā 
Refraction intro
Refraction introRefraction intro
Refraction intro
Ā 
Case study ace tablets
Case study ace tabletsCase study ace tablets
Case study ace tablets
Ā 
Dolphin System Selling Points
Dolphin System Selling PointsDolphin System Selling Points
Dolphin System Selling Points
Ā 
punya iroh suniroh Metode Pengajaran Paket C Data Anova
punya iroh suniroh Metode Pengajaran Paket C Data Anovapunya iroh suniroh Metode Pengajaran Paket C Data Anova
punya iroh suniroh Metode Pengajaran Paket C Data Anova
Ā 
Clock Jitter and Measurement
Clock Jitter and MeasurementClock Jitter and Measurement
Clock Jitter and Measurement
Ā 

Viewers also liked

Version spaces
Version spacesVersion spaces
Version spacesGekkietje
Ā 
ē¶²č·Æ安å…Ø
ē¶²č·Æ安å…Øē¶²č·Æ安å…Ø
ē¶²č·Æ安å…Øbruce761207
Ā 
Unified in Learning ā€“Separated by Space
Unified in Learning ā€“Separated by SpaceUnified in Learning ā€“Separated by Space
Unified in Learning ā€“Separated by SpaceMartin Rehm
Ā 
RCD CDHO - EE in HOME Workshop
RCD CDHO - EE in HOME WorkshopRCD CDHO - EE in HOME Workshop
RCD CDHO - EE in HOME WorkshopICF_HCD
Ā 
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...Martin Rehm
Ā 
Internet Security
Internet SecurityInternet Security
Internet Securitybruce761207
Ā 
Egitimler
EgitimlerEgitimler
Egitimleranttab
Ā 
Improve your Web Development using Visual Studio 2010
Improve your Web Development using Visual Studio 2010Improve your Web Development using Visual Studio 2010
Improve your Web Development using Visual Studio 2010Suthep Sangvirotjanaphat
Ā 
CamiƱo InglƩs en Vilar Do Colo
CamiƱo InglƩs en Vilar Do ColoCamiƱo InglƩs en Vilar Do Colo
CamiƱo InglƩs en Vilar Do Colobngcabanas
Ā 
Mobile Showcase Moblin2
Mobile Showcase Moblin2Mobile Showcase Moblin2
Mobile Showcase Moblin2Tomas Bennich
Ā 
Building advocacy and trust: social media for engagement
Building advocacy and trust: social media for engagementBuilding advocacy and trust: social media for engagement
Building advocacy and trust: social media for engagementLis Parcell
Ā 
Chemistryfm
ChemistryfmChemistryfm
ChemistryfmJoss Winn
Ā 
Presentatie Octrooigilde
Presentatie OctrooigildePresentatie Octrooigilde
Presentatie OctrooigildeMarleen Kuiper
Ā 
No Sql Introduction
No Sql IntroductionNo Sql Introduction
No Sql IntroductionDingding Ye
Ā 
The Front End Testing Frontier - RubyConf 2010
The Front End Testing Frontier - RubyConf 2010The Front End Testing Frontier - RubyConf 2010
The Front End Testing Frontier - RubyConf 2010CJ Kihlbom
Ā 
WordPress: Beyond Blogging!!
WordPress: Beyond Blogging!!WordPress: Beyond Blogging!!
WordPress: Beyond Blogging!!Joss Winn
Ā 
CSS Nite in Ginza, Vol.45
CSS Nite in Ginza, Vol.45CSS Nite in Ginza, Vol.45
CSS Nite in Ginza, Vol.45Yasuo Fukuda
Ā 
Jisc RSC Wales Glyndwr 060313 v2
Jisc RSC Wales Glyndwr 060313 v2Jisc RSC Wales Glyndwr 060313 v2
Jisc RSC Wales Glyndwr 060313 v2Lis Parcell
Ā 

Viewers also liked (20)

Version spaces
Version spacesVersion spaces
Version spaces
Ā 
ē¶²č·Æ安å…Ø
ē¶²č·Æ安å…Øē¶²č·Æ安å…Ø
ē¶²č·Æ安å…Ø
Ā 
Unified in Learning ā€“Separated by Space
Unified in Learning ā€“Separated by SpaceUnified in Learning ā€“Separated by Space
Unified in Learning ā€“Separated by Space
Ā 
RCD CDHO - EE in HOME Workshop
RCD CDHO - EE in HOME WorkshopRCD CDHO - EE in HOME Workshop
RCD CDHO - EE in HOME Workshop
Ā 
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...
Case Study on a Global Learning Program (OnlineEduca 2008 Conference Proceedi...
Ā 
Internet Security
Internet SecurityInternet Security
Internet Security
Ā 
Egitimler
EgitimlerEgitimler
Egitimler
Ā 
Improve your Web Development using Visual Studio 2010
Improve your Web Development using Visual Studio 2010Improve your Web Development using Visual Studio 2010
Improve your Web Development using Visual Studio 2010
Ā 
CamiƱo InglƩs en Vilar Do Colo
CamiƱo InglƩs en Vilar Do ColoCamiƱo InglƩs en Vilar Do Colo
CamiƱo InglƩs en Vilar Do Colo
Ā 
Eemn1
Eemn1Eemn1
Eemn1
Ā 
Mobile Showcase Moblin2
Mobile Showcase Moblin2Mobile Showcase Moblin2
Mobile Showcase Moblin2
Ā 
Building advocacy and trust: social media for engagement
Building advocacy and trust: social media for engagementBuilding advocacy and trust: social media for engagement
Building advocacy and trust: social media for engagement
Ā 
Chemistryfm
ChemistryfmChemistryfm
Chemistryfm
Ā 
Presentatie Octrooigilde
Presentatie OctrooigildePresentatie Octrooigilde
Presentatie Octrooigilde
Ā 
Scaling
ScalingScaling
Scaling
Ā 
No Sql Introduction
No Sql IntroductionNo Sql Introduction
No Sql Introduction
Ā 
The Front End Testing Frontier - RubyConf 2010
The Front End Testing Frontier - RubyConf 2010The Front End Testing Frontier - RubyConf 2010
The Front End Testing Frontier - RubyConf 2010
Ā 
WordPress: Beyond Blogging!!
WordPress: Beyond Blogging!!WordPress: Beyond Blogging!!
WordPress: Beyond Blogging!!
Ā 
CSS Nite in Ginza, Vol.45
CSS Nite in Ginza, Vol.45CSS Nite in Ginza, Vol.45
CSS Nite in Ginza, Vol.45
Ā 
Jisc RSC Wales Glyndwr 060313 v2
Jisc RSC Wales Glyndwr 060313 v2Jisc RSC Wales Glyndwr 060313 v2
Jisc RSC Wales Glyndwr 060313 v2
Ā 

Similar to WCCI 2008 Tutorial on Computational Intelligence and Games, part 2 of 3

Calculation of beta
Calculation of betaCalculation of beta
Calculation of betaiipmff2
Ā 
Ch 06 financial management notes
Ch 06 financial management notesCh 06 financial management notes
Ch 06 financial management notesBabasab Patil
Ā 
Obermeyer Case Xavier Lehnhoff
Obermeyer Case Xavier LehnhoffObermeyer Case Xavier Lehnhoff
Obermeyer Case Xavier Lehnhoffxlehnhoff
Ā 
FreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetFreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetguest4ed39d
Ā 
Insurancecompaniesdeathclaimratio 100311225415-phpapp01
Insurancecompaniesdeathclaimratio 100311225415-phpapp01Insurancecompaniesdeathclaimratio 100311225415-phpapp01
Insurancecompaniesdeathclaimratio 100311225415-phpapp01suryansh1984
Ā 
Insurance Companies Death Claim Settlement Ratio
Insurance Companies Death Claim Settlement RatioInsurance Companies Death Claim Settlement Ratio
Insurance Companies Death Claim Settlement RatioThe Financial Literates
Ā 
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARK
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARKSPICE MODEL of PBYR10100 (Standard Model) in SPICE PARK
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARKTsuyoshi Horigome
Ā 
Weight & thickness
Weight & thicknessWeight & thickness
Weight & thicknesspoonmaram
Ā 
Equations cheat sheet
Equations cheat sheetEquations cheat sheet
Equations cheat sheetgenegeek
Ā 
Bfi_barcelona08
Bfi_barcelona08Bfi_barcelona08
Bfi_barcelona08Carlo Chiorri
Ā 
Normal Distribution
Normal DistributionNormal Distribution
Normal Distributiondheard3
Ā 
Results Det 3
Results   Det 3Results   Det 3
Results Det 3guest0009aa
Ā 
FreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetFreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetMaxim Dorofeev
Ā 
Model Compression
Model CompressionModel Compression
Model CompressionDarshanG13
Ā 
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARK
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARKSPICE MODEL of 1N5818 (Standard Model) in SPICE PARK
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARKTsuyoshi Horigome
Ā 
Seepage analysis final
Seepage analysis finalSeepage analysis final
Seepage analysis finalsarvannn
Ā 
9th ICCS Noordwijkerhout
9th ICCS Noordwijkerhout9th ICCS Noordwijkerhout
9th ICCS NoordwijkerhoutGerard van Westen
Ā 

Similar to WCCI 2008 Tutorial on Computational Intelligence and Games, part 2 of 3 (20)

Financial ratios
Financial ratiosFinancial ratios
Financial ratios
Ā 
Calculation of beta
Calculation of betaCalculation of beta
Calculation of beta
Ā 
Ch 06 financial management notes
Ch 06 financial management notesCh 06 financial management notes
Ch 06 financial management notes
Ā 
Obermeyer Case Xavier Lehnhoff
Obermeyer Case Xavier LehnhoffObermeyer Case Xavier Lehnhoff
Obermeyer Case Xavier Lehnhoff
Ā 
FreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetFreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheet
Ā 
Insurancecompaniesdeathclaimratio 100311225415-phpapp01
Insurancecompaniesdeathclaimratio 100311225415-phpapp01Insurancecompaniesdeathclaimratio 100311225415-phpapp01
Insurancecompaniesdeathclaimratio 100311225415-phpapp01
Ā 
Insurance Companies Death Claim Settlement Ratio
Insurance Companies Death Claim Settlement RatioInsurance Companies Death Claim Settlement Ratio
Insurance Companies Death Claim Settlement Ratio
Ā 
Khx3200ak2 2g
Khx3200ak2 2gKhx3200ak2 2g
Khx3200ak2 2g
Ā 
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARK
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARKSPICE MODEL of PBYR10100 (Standard Model) in SPICE PARK
SPICE MODEL of PBYR10100 (Standard Model) in SPICE PARK
Ā 
Weight & thickness
Weight & thicknessWeight & thickness
Weight & thickness
Ā 
Equations cheat sheet
Equations cheat sheetEquations cheat sheet
Equations cheat sheet
Ā 
Sabrina olivares
Sabrina olivaresSabrina olivares
Sabrina olivares
Ā 
Bfi_barcelona08
Bfi_barcelona08Bfi_barcelona08
Bfi_barcelona08
Ā 
Normal Distribution
Normal DistributionNormal Distribution
Normal Distribution
Ā 
Results Det 3
Results   Det 3Results   Det 3
Results Det 3
Ā 
FreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheetFreakonomicsOfScrum spreadsheet
FreakonomicsOfScrum spreadsheet
Ā 
Model Compression
Model CompressionModel Compression
Model Compression
Ā 
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARK
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARKSPICE MODEL of 1N5818 (Standard Model) in SPICE PARK
SPICE MODEL of 1N5818 (Standard Model) in SPICE PARK
Ā 
Seepage analysis final
Seepage analysis finalSeepage analysis final
Seepage analysis final
Ā 
9th ICCS Noordwijkerhout
9th ICCS Noordwijkerhout9th ICCS Noordwijkerhout
9th ICCS Noordwijkerhout
Ā 

Recently uploaded

08448380779 Call Girls In IIT Women Seeking Men
08448380779 Call Girls In IIT Women Seeking Men08448380779 Call Girls In IIT Women Seeking Men
08448380779 Call Girls In IIT Women Seeking MenDelhi Call girls
Ā 
Top Call Girls In Jankipuram ( Lucknow ) šŸ” 8923113531 šŸ” Cash Payment
Top Call Girls In Jankipuram ( Lucknow  ) šŸ” 8923113531 šŸ”  Cash PaymentTop Call Girls In Jankipuram ( Lucknow  ) šŸ” 8923113531 šŸ”  Cash Payment
Top Call Girls In Jankipuram ( Lucknow ) šŸ” 8923113531 šŸ” Cash Paymentanilsa9823
Ā 
( Sports training) All topic (MCQs).pptx
( Sports training) All topic (MCQs).pptx( Sports training) All topic (MCQs).pptx
( Sports training) All topic (MCQs).pptxParshotamGupta1
Ā 
šŸ”|97111ą¼’99012šŸ” Call Girls In {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...
šŸ”|97111ą¼’99012šŸ” Call Girls In  {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...šŸ”|97111ą¼’99012šŸ” Call Girls In  {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...
šŸ”|97111ą¼’99012šŸ” Call Girls In {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...Diya Sharma
Ā 
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docx
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docxAlbania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docx
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docxWorld Wide Tickets And Hospitality
Ā 
08448380779 Call Girls In Karol Bagh Women Seeking Men
08448380779 Call Girls In Karol Bagh Women Seeking Men08448380779 Call Girls In Karol Bagh Women Seeking Men
08448380779 Call Girls In Karol Bagh Women Seeking MenDelhi Call girls
Ā 
JORNADA 5 LIGA MURO 2024INSUGURACION.pdf
JORNADA 5 LIGA MURO 2024INSUGURACION.pdfJORNADA 5 LIGA MURO 2024INSUGURACION.pdf
JORNADA 5 LIGA MURO 2024INSUGURACION.pdfArturo Pacheco Alvarez
Ā 
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts service
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts serviceChennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts service
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts servicevipmodelshub1
Ā 
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹atducpo
Ā 
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹å¤§å­¦å­¦ä½åŠžē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹atducpo
Ā 
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...World Wide Tickets And Hospitality
Ā 
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...baharayali
Ā 
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual service
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual serviceCALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual service
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual serviceanilsa9823
Ā 
08448380779 Call Girls In Lajpat Nagar Women Seeking Men
08448380779 Call Girls In Lajpat Nagar Women Seeking Men08448380779 Call Girls In Lajpat Nagar Women Seeking Men
08448380779 Call Girls In Lajpat Nagar Women Seeking MenDelhi Call girls
Ā 
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service šŸ¦ŗ
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service  šŸ¦ŗCALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service  šŸ¦ŗ
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service šŸ¦ŗanilsa9823
Ā 
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...Neil Horowitz
Ā 
08448380779 Call Girls In International Airport Women Seeking Men
08448380779 Call Girls In International Airport Women Seeking Men08448380779 Call Girls In International Airport Women Seeking Men
08448380779 Call Girls In International Airport Women Seeking MenDelhi Call girls
Ā 

Recently uploaded (20)

08448380779 Call Girls In IIT Women Seeking Men
08448380779 Call Girls In IIT Women Seeking Men08448380779 Call Girls In IIT Women Seeking Men
08448380779 Call Girls In IIT Women Seeking Men
Ā 
Top Call Girls In Jankipuram ( Lucknow ) šŸ” 8923113531 šŸ” Cash Payment
Top Call Girls In Jankipuram ( Lucknow  ) šŸ” 8923113531 šŸ”  Cash PaymentTop Call Girls In Jankipuram ( Lucknow  ) šŸ” 8923113531 šŸ”  Cash Payment
Top Call Girls In Jankipuram ( Lucknow ) šŸ” 8923113531 šŸ” Cash Payment
Ā 
( Sports training) All topic (MCQs).pptx
( Sports training) All topic (MCQs).pptx( Sports training) All topic (MCQs).pptx
( Sports training) All topic (MCQs).pptx
Ā 
Call Girls Service Noida Extension @9999965857 Delhi šŸ«¦ No Advance VVIP šŸŽ SER...
Call Girls Service Noida Extension @9999965857 Delhi šŸ«¦ No Advance  VVIP šŸŽ SER...Call Girls Service Noida Extension @9999965857 Delhi šŸ«¦ No Advance  VVIP šŸŽ SER...
Call Girls Service Noida Extension @9999965857 Delhi šŸ«¦ No Advance VVIP šŸŽ SER...
Ā 
šŸ”|97111ą¼’99012šŸ” Call Girls In {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...
šŸ”|97111ą¼’99012šŸ” Call Girls In  {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...šŸ”|97111ą¼’99012šŸ” Call Girls In  {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...
šŸ”|97111ą¼’99012šŸ” Call Girls In {Delhi} Cr Park ā‚¹5.5k Cash Payment With Room De...
Ā 
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docx
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docxAlbania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docx
Albania Vs Spain Albania is Loaded with Defensive Talent on their Roster.docx
Ā 
08448380779 Call Girls In Karol Bagh Women Seeking Men
08448380779 Call Girls In Karol Bagh Women Seeking Men08448380779 Call Girls In Karol Bagh Women Seeking Men
08448380779 Call Girls In Karol Bagh Women Seeking Men
Ā 
Call Girls šŸ«¤ Paharganj āž”ļø 9999965857 āž”ļø Delhi šŸ«¦ Russian Escorts FULL ENJOY
Call Girls šŸ«¤ Paharganj āž”ļø 9999965857  āž”ļø Delhi šŸ«¦  Russian Escorts FULL ENJOYCall Girls šŸ«¤ Paharganj āž”ļø 9999965857  āž”ļø Delhi šŸ«¦  Russian Escorts FULL ENJOY
Call Girls šŸ«¤ Paharganj āž”ļø 9999965857 āž”ļø Delhi šŸ«¦ Russian Escorts FULL ENJOY
Ā 
JORNADA 5 LIGA MURO 2024INSUGURACION.pdf
JORNADA 5 LIGA MURO 2024INSUGURACION.pdfJORNADA 5 LIGA MURO 2024INSUGURACION.pdf
JORNADA 5 LIGA MURO 2024INSUGURACION.pdf
Ā 
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts service
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts serviceChennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts service
Chennai Call Girls Anna Nagar Phone šŸ† 8250192130 šŸ‘… celebrity escorts service
Ā 
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
å¤§å­¦å‡ę–‡å‡­ć€ŠåŽŸē‰ˆč‹±å›½Imperialę–‡å‡­ć€‹åøå›½ē†å·„学院ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
Ā 
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹å¤§å­¦å­¦ä½åŠžē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
大学学位办ē†ć€ŠåŽŸē‰ˆē¾Žå›½USD学位čÆä¹¦ć€‹åœ£åœ°äŗšå“„大学ęƕäøščÆåˆ¶ä½œęˆē»©å•äæ®ę”¹
Ā 
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...
Spain Vs Italy 20 players confirmed for Spain's Euro 2024 squad, and three po...
Ā 
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...
Asli Kala jadu, Black magic specialist in Pakistan Or Kala jadu expert in Egy...
Ā 
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual service
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual serviceCALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual service
CALL ON āž„8923113531 šŸ”Call Girls Chinhat Lucknow best sexual service
Ā 
08448380779 Call Girls In Lajpat Nagar Women Seeking Men
08448380779 Call Girls In Lajpat Nagar Women Seeking Men08448380779 Call Girls In Lajpat Nagar Women Seeking Men
08448380779 Call Girls In Lajpat Nagar Women Seeking Men
Ā 
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service šŸ¦ŗ
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service  šŸ¦ŗCALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service  šŸ¦ŗ
CALL ON āž„8923113531 šŸ”Call Girls Saharaganj Lucknow best Female service šŸ¦ŗ
Ā 
Call Girls In Vasundhara šŸ“± 9999965857 šŸ¤© Delhi šŸ«¦ HOT AND SEXY VVIP šŸŽ SERVICE
Call Girls In Vasundhara šŸ“±  9999965857  šŸ¤© Delhi šŸ«¦ HOT AND SEXY VVIP šŸŽ SERVICECall Girls In Vasundhara šŸ“±  9999965857  šŸ¤© Delhi šŸ«¦ HOT AND SEXY VVIP šŸŽ SERVICE
Call Girls In Vasundhara šŸ“± 9999965857 šŸ¤© Delhi šŸ«¦ HOT AND SEXY VVIP šŸŽ SERVICE
Ā 
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...
Atlanta Dream Exec Dan Gadd on Driving Fan Engagement and Growth, Serving the...
Ā 
08448380779 Call Girls In International Airport Women Seeking Men
08448380779 Call Girls In International Airport Women Seeking Men08448380779 Call Girls In International Airport Women Seeking Men
08448380779 Call Girls In International Airport Women Seeking Men
Ā 

WCCI 2008 Tutorial on Computational Intelligence and Games, part 2 of 3

  • 1. CIG case study: car racing ā€¢ A prolonged example of applying CI to a game: car racing ā€¢ Sensor representation and input selection ā€¢ Incremental evolution ā€¢ Competitive coevolution ā€¢ Player modelling ā€¢ Content creation
  • 2. Racing games ā€¢ On the charts for the last three decades ā€¢ Can be technically simple (computationally cheap) or very sophisticated ā€¢ Easy to pick up and play, but possess almost unlimited ā€œdepthā€ (a lifetime to master) ā€¢ Can be played on your own or with others
  • 3. CI in racing games ā€¢ Learning to race ā€¢ on your own, against speciļ¬c opponents, against opponents in general, on one or several tracks, using simple or complex cars/physics models, etc. ā€¢ Modelling driving styles ā€¢ Creating entertaining game content: tracks and opponent drivers
  • 4. A simple car game ā€¢ Optimised for speed, not for prettiness ā€¢ 2D dynamics (momentum, understeer, etc.) ā€¢ Intended to qualitatively replicate a standard toy R/C car driven on a table ā€¢ Bang-bang control (9 possible commands)
  • 5. ā€¢ Walls are solid ā€¢ Waypoints must be passed in order ā€¢ Fitness: continuous approximation of waypoints passed in 700 time steps
  • 6. ā€¢ Inputs ā€¢ Six range-ļ¬nder sensors (evolvable pos.) ā€¢ Waypoint sensor, Speed, Bias ā€¢ Networks ā€¢ Standard MLP, 9:6:2 ā€¢ Outputs interpreted as thrust/steering
  • 7. T rack 10 50 100 200 P r. 1 1.9 (0.1) 1.99 (0.06) 2.02 (0.01) 2.04 (0.02) 10 2 2.06 (0.1) 2.12 (0.04) 2.14 (0) 2.15 (0.01) 10 3 3.25 (0.08) 3.4 (0.1) 3.45 (0.12) 3.57 (0.1) 10 4 3.35 (0.11) 3.58 (0.11) 3.61 (0.1) 3.67 (0.1) 10 5 2.66 (0.13) 2.84 (0.02) 2.88 (0.06) 2.88 (0.06) 10 6 2.64 (0) 2.71 (0.08) 2.72 (0.08) 2.82 (0.1) 10 7 1.53 (0.29) 1.84 (0.13) 1.88 (0.12) 1.9 (0.09) 10 T rack 18 2 0.59 (0.15) 3 0.73 (0.22) 4 0.85 (0.21) 5 0.93 (0.25) 06 7 8 Fitness (sd) 1.66 (0.12) 1.86 (0.02) 2.27 (0.45) 2.66 (0.3) TABLE VI 2.19 (0.23) 2.47 (0.18) 0.22 (0.15) 0.15 (0.01) TABLE V F ITNESS OF BEST CONTROLLERS , EVOLVING CONTROLLERS F ITNESS OF A FURTHER EVOLVED GENERAL CONTROLLER WITH EVOLVABLE SENSOR PARAMETERS ON THE DIFFERENT TRACKS . C OMPOUND FITNESS SPECIALISED FOR EACH TRACK , STARTING FROM A FURTHER EVOLVED 2.22 (0.09). GENERAL CONTROLLER WITH EVOLVED SENSOR PARAMETERS . Fig. 6. Sensor setup of a controller reach good ļ¬tness on, track 7. Presum their angular spread reļ¬‚ects the larg T rack 10 50 100 200 P r. T rack 10 50 100 200 Phas to handle in order to navigate th r. 1 1.9 (0.1) 1.99 (0.06) 2.02 (0.01) 1 (0.02) 2.040.32 (0.07) 100.54 (0.2) 0.7 (0.38) 0.81 (0.5) 2 2 2.06 (0.1) 2.12 (0.04) 2.14 (0) 2 2.150.38 (0.24) (0.01) 100.49 (0.38) 0.56 (0.36) 0.71 (0.5) 2 3 3.25 (0.08) 3.4 (0.1) 3.45 (0.12) 3 (0.1) 3.570.32 (0.09) 100.97 (0.5) 1.47 (0.63) 1.98 (0.66) 7 4 3.35 (0.11) 3.58 (0.11) 3.61 (0.1) 4 (0.1) 3.670.53 (0.17) 101.3 (0.48) 1.5 (0.54) 2.33 (0.59) 9 5 2.66 (0.13) 2.84 (0.02) 2.88 5 (0.06) 2.880.45 (0.08) (0.06) 100.95 (0.6) 0.95 (0.58) 1.65 (0.45) 8 6 2.64 (0) 2.71 (0.08) 2.72 6 (0.08) 0.4 (0.08) 2.82 (0.1) 100.68 (0.27) 1.02 (0.74) 1.29 (0.76) 5 7 1.53 (0.29) 1.84 (0.13) 1.88 7 (0.12) 1.9 0.3 (0.07) (0.09) 100.35 (0.05) 0.39 (0.09) 0.46 (0.13) 0 8 0.16 (0.02) 0.19 (0.03) 0.2 (0.01) 0.2 (0.01) 0 8 0.59 (0.15) 0.73 (0.22) 0.85 (0.21) 0.93 (0.25) 0 TABLE I TABLE VI T HE FITNESS OF THE BEST CONTROLLER OF VARIOUS GENERATIONS ON F ITNESS OF BEST CONTROLLERS , EVOLVING CONTROLLERS TRACKS , AND NUMBER OF RUNS PRODUCING THE DIFFERENT SPECIALISED FOR EACH TRACK , STARTING FROMPROFICIENT CONTROLLERS . F ITNESS AVERAGED OVER 10 SEPARATE A FURTHER EVOLVED GENERAL CONTROLLER WITH EVOLVED SENSOR PARAMETERS . STANDARD DEVIATION BETWEEN PARENTHESES . EVOLUTIONARY RUNS ; Fig. 2. The initial sensor setup, which is kept throughout the evolutionary Fig. 6. 5. track Sensor setup of a controller specialized for, and able to consistently run for those runs where sensor parameters are not evolvable. Here, setup of controller specialized forreach good While on, track 7. Presumably the use of all but one sensor and Fig. 5. Sensor the car more or less retaining the two longest-range sensors from the further evolved ļ¬tness general is seen in close-up moving upward-leftward. At this particular position, the their angular spread reļ¬‚ects the large variety of different situations the car front-right sensor returns a positive number very close to 0, as it detects on, it has added medium-range sensors in the front and controller it is based a has to handle in order to navigate this more difļ¬cult track. wall near the limit of its range; the front-left sensor returns a number close back,The front, very short-range sensor to the left. number of waypoints in the track, 7. Sensor setup of another con to 0.5, and the back sensor a slightly larger number. and a left and right passed, divided by the Fig. sensors do not detect any walls at all and thus return 0. plus an intermediate term representing how far it is on its way in ļ¬gure 6 seemingly using all i one to the next waypoint, calculated from the relative distances between the car and the previous and next waypoint. A range 200 pixels, as has three sensors pointing forward- ļ¬tness of 10 evolutionary runs were made, track controllers. For each track, 1.0 thus means having completed one full VII. O BSERVATIONS ON EV left, forward-right and backward respectively. The two other within the alloted time. Waypoints can only be passed in the where the initial population was seeded with the general sensors, which point left and right, have reach 100; this is correct order, and a waypoint is counted as passed when the illustrated in ļ¬gure 2. controller and evolution of the car is within 30continue for waypoint. In It has previously been found centre was allowed to pixels from the 200
  • 8.
  • 9. Example video Evolved with 50+50 ES, 100 Generatons
  • 10. Choose your inputs (+their representation) ā€¢ Using third-person inputs (cartesian inputs) seems not to work ā€¢ Either range-ļ¬nders or waypoint sensor can be taken away, but some ļ¬tness lost ā€¢ A little bit of noise is not a problem, actually itā€™s desirable ā€¢ Adding extra inputs (while keeping core inputs) can reduce evolvability drastically!
  • 11. If you donā€™t know your inputs... ā€¢ Memetic techniques (e.g. memetic ES) can sort out useful from useless inputs ā€¢ Principle: evolve neural network weights together with a mask: whether connections are on or off ā€¢ Masks and weights are evolved at different time scales; after every mask mutation, weight space is searched - if no ļ¬tness increase, the mask is reverted
  • 12.
  • 13. Learning controllers with irrelevant inputs present Togelius, Gomez and Schmidhuber (2008)
  • 14. Generalization and specialization ā€¢ A controller evolved for one track does not necessarily perform well on other tracks ā€¢ How do we achieve more general game- playing skills? ā€¢ Is there a tradeoff between generality and performance?
  • 15. damaging such cars in collisions is ha weight. The dynamics of the car are based on mechanical model, taking into account car and bad grip on the surface, but is n measurement [13][14]. The model is s [4], and differs mainly in its improve after more experience with the physical response system was reimplemented to realistic (and, as an effect, more undesir may cause the car to get stuck if the unfortunate angle, something often see physical cars. A track consists of a set of walls, a and a set of starting positions and di is added to a track in one of the sta corresponding starting direction, both t being subject to random alterations. Th for ļ¬tness calculations. For the experiments we have des tracks, presented in ļ¬gure 1. The tr vary in difļ¬culty, from easy to hard. are versions of three other tracks wi in reverse order, and the directions of reversed. The main differences between our real R/C car racing problem have to reported in Tanev et al. as well as [4] not unimportant lag in the communica computer and car, leading to the control perceptions. Apart from that, there Fig. 1. The eight tracks. Notice how tracks 1 and 2 (at the top), 3 and 4, 5 and 6 differ in the clockwise/anti-clockwise layout of waypoints and in estimations of the carā€™s position a associated starting points. Tracks 7 and 8 have no relation to each other overhead camera. In contrast, the sim
  • 16. Incremental evolution ā€¢ Introduced by Gomez & Mikkulainen (1997) ā€¢ Change the ļ¬tness function f (to make it more demanding) as soon as a certain ļ¬tness is achieved ā€¢ In this case, add new tracks to f as soon as the controller can drive 1.5 rounds on all tracks currently in f
  • 18. ā€¢ Controllers evolved for speciļ¬c tracks perform poorly on other tracks ā€¢ General controllers, that can drive almost any track, can be incrementally evolved ā€¢ Starting from a general controller, a controller can be further evolved for specialization on a particular track ā€¢ drive faster than the general controller ā€¢ works even when evolution from scratch did not work!
  • 19. Two cars on a track ā€¢ Two car with solo-evolved controllers on one track: disaster ā€¢ they donā€™t even see each other! ā€¢ How do we train controllers that take other drivers into account? (avoiding collisions or using them to their advantage) ā€¢ Solution: car sensors (rangeļ¬nders, like the wall sensors) and competitive coevolution
  • 21. Competitive coevolution ā€¢ The ļ¬tness function evaluates at least two individuals ā€¢ One individualā€™s success is adversely affected by the otherā€™s (directly or indirectly) ā€¢ Very potent, but seldom straightforward; e.g. Hillis (1991), Rosin and Belew (1996)
  • 22. Competitive coevolution ā€¢ Standard 15+15 ES; each individual is evaluated through testing against the current best individual in the population ā€¢ Fitness function a mix of... ā€¢ Absolute ļ¬tness: progress in n time steps ā€¢ Relative ļ¬tness: distance ahead of or behind the other car after n time steps
  • 26. Problems with coevolution ā€¢ Over-specialization and cycling ā€¢ Can be battled with e.g. archives ā€¢ Loss of gradient ā€¢ Can be battled with careful ļ¬tness function design, e.g. combining absolute and relative ļ¬tness ā€¢ Much more research needed here!
  • 27. Multi-population coevolution ā€¢ Typically, competitive coevolution uses one or two populations ā€¢ Many more populations can be used! ā€¢ Can help against cycling and overspecialization ā€¢ The phenotypical diversity between populations can be useful in itself
  • 28. Example: 1 versus 9 populations Togelius, Burrow, Lucas (2007)
  • 29. Player modelling ā€¢ Can we create players that drive just like speciļ¬c human players? ā€¢ The models need to be... ā€¢ Similar in terms of performance ā€¢ Similar in terms playing (driving) style ā€¢ Robust
  • 30. Direct modelling ā€¢ Let a player drive a number of tracks ā€¢ Use supervised learning to associate inputs (sensors) with outputs (driving commands) ā€¢ e.g. MLP/Backpropagation or k-nearest neighbour ā€¢ Suffers from generalization problems, and that any approximation is likely to lead to worse playing performance
  • 31. Indirect modelling ā€¢ Let a human drive a test track, record performance, speed and orthogonal deviation at the various waypoints the track ā€¢ Start from a good, general evolved neural network controller, and evolve it further ā€¢ Fitness: negative difference between controller and player for the three measures above
  • 32. The test track supposedly requires a varied repertoire of driving skills 1 0.8 0.6 Fitness (progess, speed) 0.4 0.2 0 āˆ’0.2 āˆ’0.4 āˆ’0.6 āˆ’0.8 0 0 10 Fig. 2. The test track and the car. Fig. 3. Evolving a First of all, we design a test track, featuring a number of different types of racing challenges. The track, as pictured 0 in (ļ¬g), has two long straight sections where the player can āˆ’0.2
  • 33. Content creation ā€¢ Creating interesting, enjoyable levels, worlds, tracks, opponents etc. ā€¢ Not the same as well-playing opponents ā€¢ Probably the area where commercial game developers need most help ā€¢ What makes game content fun? Many theories, e.g. Thomas Malone, Raph Koster, MihĆ”ly CsĆ­kszentmihĆ”lyi
  • 34. Track evolution ā€¢ Using the controllers we evolved to model human players, we evolve tracks that are fun to drive for the modelled player ā€¢ Fitness function: ā€¢ Right amount of progress ā€¢ Variation in progress ā€¢ High maximum speed
  • 35. Fig. 5. Track evolved using the random walk initialisation and mutation. e the representa- nted with several t the beginning plementations of nļ¬gurations are rd initial track rners. Each mu- ontrol points by distribution with y axes. xperiments, mu- onļ¬guration, but ectangle track is eds of mutations those mutations controller is not e result of such a ll drivable track. ck and evolution Fig. 6. A track evolved (using the radial method) to be fun for the ļ¬rst author, who plays too many racing games anyway. It is not easy to drive, which is just as it should be. n, starts from an rol points around
  • 36. the results of ou car racing [10]. In the section describe a numb value, most of wh described here. D sures would deļ¬n urgent to study th oft-cited hypothe know there are n entertainment me games and types needed. Finally we no different approach in the beginning Fig. 7. A track evolved (using the radial method) to be fun for the second viewed from sev author, who is a bit more careful in his driving. Note the absence of sharp on using evoluti turns. in games is not studying under w perspective we h
  • 37. ks by sampling aken advantage ack. First thick side of the b- ixels or subject nt is set up. But th of the track, and sometimes struction of the ing the b-spline middle of the imately regular e resulting track est track which e control points Fig. 5. Track evolved using the random walk initialisation and mutation. the representa- ed with several the beginning
  • 38. but only sometimes causes the car to collide. Those elements are believed to be the main source of ļ¬nal progress variability. These features are also notably absent from track c, on which the good player model has very low variability. The progress of the controller is instead limited by many broad curves. Fig. 3. Three evolved tracks: ((a)) evolved for a bad player with target progress 1.1, (b) evolved for a good player with target ļ¬tness 1.5, (c) evolved for a good player with target progress 1.5 using only progress ļ¬tness.
  • 40. Video: real car control
  • 41. More on these topics ā€¢ http://julian.togelius.com ā€¢ e.g. Togelius, Lucas and De Nardi: ā€œComputational Intelligence in Racing Gamesā€ ā€¢ Togelius, Gomez and Schmidhuber: ā€œLearning what to ignoreā€ on Friday, 11.10, room 606 ā€¢ Car Racing Competition on Tuesday 15.00, room 402