AIMA: Chapter 2 Choueiry In tro dution to Artiial In - - PowerPoint PPT Presentation

▶
aima chapter 2 choueiry in tro du tion to arti ial in
SMART_READER_LITE
LIVE PREVIEW

AIMA: Chapter 2 Choueiry In tro dution to Artiial In - - PowerPoint PPT Presentation

Aug 03, 2023 531 likes •757 views

B.Y. Title: In telligen t Agen ts AIMA: Chapter 2 Choueiry In tro dution to Artiial In telligene CSCE 476-876, Spring 2016 URL: www.se.unl.edu/~ ho ue iry /S1 6- 476 -87 6 1 Berthe Y. Choueiry (Sh

slide-1
SLIDE 1

✬ ✫ ✩ ✪

Title: In telligen t Agen ts AIMA: Chapter 2 In tro du tion to Arti ial In telligen e CSCE 476-876, Spring 2016 URL:
  • www. se.unl.edu/~
ho ue iry /S1 6- 476
  • 87
6 Berthe Y. Choueiry (Sh u-w e-ri) (402)472-5444 B.Y. Choueiry 1 Instru tor's notes #4 Jan uary 25, 2016
slide-2
SLIDE 2

✬ ✫ ✩ ✪

In telligen t Agen ts 1. Agen ts and en vironmen ts 2. Rationalit y 3. PEAS Sp e ifying the task en vironmen t: P erforman e measure, En vironmen t, A tuators, Sensors 4. T yp es
  • f
en vironmen ts 5. T yp es
  • f
In telligen t Agen ts B.Y. Choueiry 2 Instru tor's notes #4 Jan uary 25, 2016
slide-3
SLIDE 3

✬ ✫ ✩ ✪

Agen t An ything that

  

p er eiv es its en vironmen t through sensors a ts up
  • n
its en vironmen t through a tuators Agen ts in lude: Humans, rob
  • ts,
soft w are, et . Sensors? A tuators? The agen t fun tion maps from p er ept sequen es to a tions:

f : P∗ → A

The agen t program runs
  • n
the ph ysi al ar hite ture to pro du e f B.Y. Choueiry 3 Instru tor's notes #4 Jan uary 25, 2016
slide-4
SLIDE 4

✬ ✫ ✩ ✪

V a uum- leaner w
  • rld

A B

P er epts: lo ations and
  • n
ten ts, e.g., [A, dirty] A tions: Left , Right , Suck , NoOp B.Y. Choueiry 4 Instru tor's notes #4 Jan uary 25, 2016
slide-5
SLIDE 5

✬ ✫ ✩ ✪

A V a uum- leaner Agen t P er ept sequen e A tion

[A, Clean]

Righ t

[A, Dirty]

Su k

[B, Clean]

Left

[B, Dirty]

Su k

[A, Clean]

,[A, Clean] Righ t . . .

[A, Clean]

,[A, Clean] ,[A, Clean] Righ t . . . F un tion Reex-V a uum-Agen t ([location, status]]) returns an a tion if status = Dirty then return Suck else if location = A then return Right else if location = B then return Left B.Y. Choueiry 5 Instru tor's notes #4 Jan uary 25, 2016
slide-6
SLIDE 6

✬ ✫ ✩ ✪

Goal
  • f
AI Build rational agen ts. Rational = ? What is rational dep ends
  • n:
1. P erforman e measures (ho w, when) 2. The agen ts' prior kno wledge
  • f
the en vironmen t 3. The a tions the agen t an p erform 4. P er ept sequen e to date (history): ev erything agen t has p er eiv ed so far B.Y. Choueiry 6 Instru tor's notes #4 Jan uary 25, 2016
slide-7
SLIDE 7

✬ ✫ ✩ ✪

P erforman e meaure Fixed p erforman e measure ev aluates the en vironmen t sequen e
  • ne
p
  • in
t p er square leaned up in time t
  • p
  • in
t p er lean square p er time step, min us
  • ne
p er mo v e?
  • p
enalize for > k dirt y squares? B.Y. Choueiry 7 Instru tor's notes #4 Jan uary 25, 2016
slide-8
SLIDE 8

✬ ✫ ✩ ✪

Rationalit y A rational agen t ho
  • ses
whi hev er a tion maximizes the exp e ted v alue
  • f
the p erforman e measure giv en the p er ept sequen e to date Rational =
  • mnis ien
t, lairv
  • y
an t Rationalit y maximizes exp e ted p erforman e P erfe tion maximizes a tual p erforman e Rational =

⇒

exploration, learning, autonom y After a su ien t exp erien e
  • f
its en vironmen t, b eha vior
  • f
a rational agen ts b e omes ee tiv ely indep enden t
  • f
prior kno wledge. B.Y. Choueiry 8 Instru tor's notes #4 Jan uary 25, 2016
slide-9
SLIDE 9

✬ ✫ ✩ ✪

PEAS T
  • design
a rational agen t, w e m ust sp e ify the task en vironmen t P erforman e measure? En vironmen t? A tuators? Sensors? Consider, e.g., the task
  • f
designing an automated taxi.. B.Y. Choueiry 9 Instru tor's notes #4 Jan uary 25, 2016
slide-10
SLIDE 10

✬ ✫ ✩ ✪

PEAS: Automated taxi P erforman e measure: safet y , destination, prots, legalit y ,
  • mfort, . . .
En vironmen t: US urban streets, freew a ys, tra , p edestrians, stra y animals, w eather, . . . A tuators: steering, a elerator, brak e, horn, sp eak er/displa y , . . . Sensors: video, a elerometers, gauges, engine sensors, k eyb
  • ard,
GPS, . . . B.Y. Choueiry 10 Instru tor's notes #4 Jan uary 25, 2016
slide-11
SLIDE 11

✬ ✫ ✩ ✪

En vironmen t (1) 1. F ully Observ able vs. P artially Observ able 2. Deterministi vs. sto hasti 3. Episo di vs. sequen tial 4. Stati vs. dynami 5. Dis rete vs.
  • n
tin uous 6. Single agen t vs. m ultiagen t B.Y. Choueiry 11 Instru tor's notes #4 Jan uary 25, 2016
slide-12
SLIDE 12

✬ ✫ ✩ ✪

En vironmen t (2) F ully/P artially Observ able: sensors an dete t all asp e ts
  • f
the w
  • rld
Ee tiv ely fully
  • bserv
able: relev an t asp e ts Deterministi vs. sto hasti : from the agen t's view p
  • in
t Next state determined b y urren t state and agen ts' a tions P artially
  • bserv
able + deterministi app ears sto hasti Episo di vs. sequen tial: Agen t's exp erien e divided in to atomi episo des; subsequen t episo des do not dep end
  • n
a tions in previous episo des B.Y. Choueiry 12 Instru tor's notes #4 Jan uary 25, 2016
slide-13
SLIDE 13

✬ ✫ ✩ ✪

En vironmen t (3) Stati vs. dynami : Dynami : En vironmen t hanges while agen t is delib erating Semidynami : en vironmen t stati , p erforman e s ores dynami Dis rete vs.
  • n
tin uous: Finite n um b er
  • f
pre epts, a tions Single agen t vs. m ultiagen t: B 's b eha vior maximizes a p erforman e measure whose v alue dep ends
  • n A's
b eha vior. Co
  • p
erativ e,
  • mp
etitiv e,
  • mm
uni ation. Chess? T axi driving? hardest ase? B.Y. Choueiry 13 Instru tor's notes #4 Jan uary 25, 2016
slide-14
SLIDE 14

✬ ✫ ✩ ✪

En vironmen t (4) Hardest ase: patially
  • bserv
able, sto hasti , sequen tial, dynami ,
  • n
tin uous, and m ultiagen t Solitaire Ba kgammon In ternet shopping T axi Observ able Deterministi Episo di Stati Dis rete Single-agen t Answ ers dep end
  • n
ho w y
  • u
dene/in terpret the ase Episo di : hess tournamen t B.Y. Choueiry 14 Instru tor's notes #4 Jan uary 25, 2016
slide-15
SLIDE 15

✬ ✫ ✩ ✪

En vironmen t t yp es Solitaire Ba kgammon In ternet shopping T axi Observ able Y es Y es No No Deterministi Y es No P artly No Episo di No No No No Stati Y es Semi Semi No Dis rete Y es Y es Y es No Single-agen t Y es No Y es No (ex ept au tions) The en vironmen t t yp e largely determines the agen t design The real w
  • rld
is (of
  • urse)
partially
  • bserv
able, sto hasti , sequen tial, dynami ,
  • n
tin uous, m ulti-agen t B.Y. Choueiry 15 Instru tor's notes #4 Jan uary 25, 2016
slide-16
SLIDE 16

✬ ✫ ✩ ✪

T yp es
  • f
Agen ts F
  • ur,
in
  • rder
  • f
in reasing generalit y: 1. Simple reex agen ts 2. Simple reex agen ts with state 3. Goal-based agen ts 4. Utilit y-based agen ts 5. Learning agen ts All these an b e turned in to learning agen ts. B.Y. Choueiry 16 Instru tor's notes #4 Jan uary 25, 2016
slide-17
SLIDE 17

✬ ✫ ✩ ✪

Simple reex agen ts
  • Simple
lo
  • k-up
table, mapping p er epts to a tions, is
  • ut
  • f
question (to
  • large,
to
  • exp
ensiv e to build)
  • Man
y situations an b e summarized b y
  • ndition-a tion
rules (h umans: learned resp
  • nses,
innate reexes)

Agent Environment

Sensors What action I should do now Condition-action rules Actuators What the world is like now

Re tangles: agen t's in ternal state Ov als: ba kground information Implemen tation: easy; Appli abilit y: narro w B.Y. Choueiry 17 Instru tor's notes #4 Jan uary 25, 2016
slide-18
SLIDE 18

✬ ✫ ✩ ✪

Simple reex agen ts with state
  • Sensory
information alone is not su ien t
  • Need
to k eep tra k
  • f
ho w the w
  • rld
ev
  • lv
es (ev
  • lution:
indep enden tly
  • f
agen t,
  • r
aused b y agen t's a tions)

Agent Environment

Sensors State How the world evolves What my actions do Condition-action rules Actuators What the world is like now What action I should do now

Ho w the w
  • rld
ev
  • lv
ed: mo del-based agen t B.Y. Choueiry 18 Instru tor's notes #4 Jan uary 25, 2016
slide-19
SLIDE 19

✬ ✫ ✩ ✪

Goal-based agen ts
  • State
& a tions don't tell where to go
  • Need
goals to build sequen es
  • f
a tions (planning)

Agent Environment

Sensors What action I should do now State How the world evolves What my actions do Actuators What the world is like now What it will be like if I do action A Goals

Goal-based: uses the same rules for dieren t goals Reex: will need a
  • mplete
set
  • f
rules for ea h goal B.Y. Choueiry 19 Instru tor's notes #4 Jan uary 25, 2016
slide-20
SLIDE 20

✬ ✫ ✩ ✪

Utilit y-based agen ts
  • Sev
eral a tion sequen es to a hiev e some goal (binary pro ess)
  • Need
to sele t among a tions & sequen es. Preferen es.
  • Utilit
y: State → real n um b er (express degree
  • f
satisfa tion, sp e ify trade-os b et w een
  • ni ting
goal)

Agent Environment

Sensors How happy I will be in such a state State How the world evolves What my actions do Utility Actuators

  • What action I

should do now What it will be like if I do action A What the world is like now

B.Y. Choueiry 20 Instru tor's notes #4 Jan uary 25, 2016
slide-21
SLIDE 21

✬ ✫ ✩ ✪

Learning agen ts Agen t
  • p
erates in an initially unkno wn en vironmen t, and b e omes more
  • mp
eten t than its initial kno wledge alone migh t allo w

Performance standard

Agent Environment

Sensors Performance element changes knowledge learning goals Problem generator feedback Learning element Critic Actuators

  • Learning:
pro ess
  • f
mo di ation
  • f
ea h
  • mp
  • nen
t
  • f
the agen t to bring the
  • mp
  • nen
ts in to loser agreemen t with the a v ailable feedba k information, th us impro ving
  • v
erall p erforman e
  • f
the agen t. B.Y. Choueiry 21 Instru tor's notes #4 Jan uary 25, 2016