Scribd
Upload
17 views150 pages

AI Notes

Uploaded by

Punya Nair
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
17 views150 pages

AI Notes

Uploaded by

Punya Nair
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
Moacdole +. . _5_ mk 5 Module 2 5-to' mes max l 7 Module 2 lo-I§_mles “Module yb O-S meg «| Moolulo 6 &-10 mks + Q0-25 mes i BY to-'€ mk . o 10 mes .) _ 10 lo mks “FB -AS Lu YO mes | Nes | mel Uhah is Act ¢ | rteltigence. is the ability 16 think. awd OLhiewe _desived goals ands .we ean sa Tndeligence is useat for PROBLEM | OUVING = Intelligence iS Said to be avnigue feature of human beings and wih 'e\n this ivelligenG is implemented in machines its called A 39 TAL is _aveg. of Computer sctence -forusing a} creating. machinec Soak can Snow toenavigu? al _simmlan7A0 humans” : ao10 la q AZ is ne sivay of how to make compote 0" do “Tnings which numdns can Ao. tay is O branch of computer science oat i al Cove rmeck wlth quem ation of wteligent behav our AL is an _art of creating machines Anort performs, anctions tok require woreiligence. usnen pe(fermed toy people. PAL is existmg new effort make compote y” Anink wit mind i full and titeral seve 8 | Al systems t= tethink like Humane - act we numMans_ etal el oleloleraral al al aal « ok {- 12M clevelo ped ches? pla ying. computer «| Appl cations asf ML . : os - Game Play called (BEEP RLVE? which defeated _ advil Champion ~ Gary Kaspatdy in 1aq7 Avionomous Planning and Scheduling =~ NASA clevéloped - «Remote Agent Program | fo control tne Scheduling of operation fon O_spacecwa ht —Remele Agent Program maadilored Sperartis oF spacesinip~ and provided info to NASAL Rutenomous Comtyol = The ALWINN Com puter visiod contre sunter was tamed 4 dyve acar tniy sgstem was Filed in migivan votin vdeo ramerd. which transmit (ad images and dvpending 00 images best cluection bs drive wis selected Diagnosis = Medical _Aiagnasis program used __ symptom anaiys\s 19 predict dise@se and gave result equivalem -b @ pect prysiclan [Ropotics |- Many surgeon use Robot called ip i Xo do” microsurgery aus fins devwleped & eaystyes Playing _ and Ze plonving. Ebel 4 Co Arvdomatice * ‘Logistics Planning and scheduling foo Aras pork otto v = This t60l Allowedk Plan to be generated” Wn hours on Providing Aart poimt and end destination. . YWWamMic oamalysis 98 ddal: ! odo auavugaoga Seanad ddaodaan e iN Al problems : _ . : a I yrypry yy Yu Al hod oral focus on (ermal tasks ke gare Playing And Ahesfem roving _Q3 these Gennes Ore cenndewd ie be cisplaxiug intelige But as tne rampenity “of Game increase ibuvas clifficult fr computer to steve Ou pass ibl Sutons wbich Con be easily-dene by The | Wuman brain Avotner focus of Al was io solve problem telate 46 commen sense (easaning Sir example ometning {alison tne floor, it yng. breaR ane te invesngate _S ype of Ceamaing , General «| Peoblem Selvet CGPS) was built 4 These machines could solve simple cammov) sens problem as Ne one tought of writing large KB proararn AAs fesearcheel progressed At system wil a] Sensocg were built vision and speech Were. | mosk wuidely Used sensors tut Pxdllem 0 seperation ce noise {rom speech a5 well a design System Wr nerlura\ Language. Yrecessiv Was crallen ging = The_pirolem of undersionding spoken Langa 4¢ [is very alifficul 4e solve fer ang Reasoning CH machines = A\_ machines inthally clesigne \ egenera) marin ¢ weve equipped usitia eacnmon sense and = Pee ptwal Ccensuns, shit because machimes wel i skit acl af AG £4 et - yeguiive upnitirated tee i j | AY technique | : an : anne 5 oy Tre Ui gence eG Uies pnisledge ous Priaste Age “Acnas features (ike :- : 0 {bre is volumiaous “5 wp 4 is hard to characterize: accurately 5 iyPTe (S comstamity changing 3 iv[ Th is organized in 4 clivferenmt wy compared, uj to lata organized (storage architecture) 5 a =o ial th ii AV a _litself efficletly. Ac dechniques is G& metnocl to explott This | pnew-ivdge poe wo KB sinicla nas: yeproemang as gixen below? knowledge tyeseak Wo € AR) must eqptuye apnerdiized belhawieur Lo similan siuattos Knowledge musttoe uy dene: people Te must be easily modifiable to adppt ts ‘the changing world T+ must pe Gble de provide approximate lute 4s many Qreat stuations . : T)_must ve able to fe pond quickly be mainlainin % 3 —a Al techniques are used Jo seve Ai peck] eme and. a sige Al pro\lem ca be solved esta multi pl AL +ecrniques A 3 imperlant Al techniques gye t [Seaveln: provide o way of sriving piovlem whene. ae ial \S GN mete gu oN 5 » a a =i = 2 i |Alostrachow: | perainon oF 2 len padant. onc less dant Pnowsledge needy ° te solve. oO pradem, : ° : i, . Ot Components of AT 5 : oO | Sensors -. ‘ 2. They actos courte of ints fran “Al Sustern 69 = 4 Camera heal detector mike radon detector oO 9 2 | Effector (Actuadtore) ; p They are components respansivle te comy oud 7 ACsived dash -©4' Robotic om’ = 2. 2 alKnowledge Base 7 Stores all “tne facls and wiley Qa g porticut 2 pidlolewn Coma . T+ makertnem avdiidyie to : witerence engine to era new Andling: = ul anferene Engine oI i The program’ “nat locades » Oppropyirle bveiwerk J ‘ : ; : 5 by _clpplving logical prove WOg-Auncl Problem shige Acte gies. o wl Me ey el elu — “9 = 0 c ° salons is cape by Tnteligent Agent? A An ogeod is aneyftaia | ‘tha, can be viewed 2) ei ex S)Ks_enmvenm ent _., thyough sensees and aeding upon the @iawivenment “Hnrougiy actuadats Crobbstic acto Tq numan Agent + SenSovs ave eye, nose ,\ipi SKIN €[Link] Actuaders are hancl lege mouth ele Tn Robohc OYeNe: Sensors are camera Terme infcoved saves & Ockrdtors Are Robotic arr) maby 3 Peveephen (sense) Sere pO €) 5 i Envifonmen ty St © puadon 5 - a Oa abe, getferm acne 5 Agent = Prenilecture Cru) + Progvang Cfunctioy) } ; Nac - Nai 3 . Lt T a 1 ' { d ~ ; a 0 ~ qd Action ~ eve agirt _ ‘qaen: _ Aas perceive S calied €°T SEQUENCE - Mapping of Percept. sequence to ain ac hon done_by AGENT SANCTION cS ~ AGLVT Function can ofificigh agent | & implememed by AdeNtT P2oGeAM oe AGENT FUNC Mon are implemented hey ACEC - PROGRAM winch Tum Cemmbed ded) an fGen. AgCH MEOLRE. fp Fer _ecicln pos n possible —_pertey periepl an ideat vats fonal 3 agent should do usnatever aethon is eR PEle to__maximize its performance “Medsure a “ne, basis BF evidence brovihecd oy tne perce pt Seqaeyee ancl the bu le in Puscebed ge tes, ayenk nas ‘ Rahevality maximizes expecteck petformance Example + & vorryenal vocuum Meaneragent has features given belo: - Sf n Au| Poimt quwarded for acl clean SqAra re Lut Cleography /geormedyy of environment is know A\prioc Fir acvance) 3] The ditt olistibution ond tathal locaton of agent is UMPy ovo 4 | Operations possvolé are : Let move Pigt’ maye al Rrword mae , Dathwora mere. Suck arocl idle. S |Perdewve | [cahon £ amour of Aith & | MWe agent usi\\ become irpationa) if after Cleaning 1s done tt SHA oscillates 1eC4 - rsh Ond ‘ne re iS penally for €ach Mave. si\BeHey aaene will be tclle ance all Sauares axe ct S| Agent must cnecla Wcasiqnally if avee needed _ to ye-Clecin Pediadreally : Qi re qeanelys} is_Unbnden then thas to ereplore.. 7 | Br aclean area. : 3 a a 3 m Ts a ed & te | Oraniscent “Agent = — pA aqent _ulho bows ee Bald Pame of peal aokten avid cain ace aceeraling ty but th 9. agent Ore impossile te clesign. - : Ate we cap design an ageot “Inet classed Oo Aums. out best if unusua! Wings lnappen . ~ Such agent are clifficult to hesign_avich 3 | Ove colllecl Cnoniscent Agente Example: 2 Suppose an agend fod sud vodel ix empty an ert essing Toack bud in midale Roy ¢ | Container Moning 60 Stéy walk Falls on agen Ond agent is Matnw iF ue can design ager Je ingndie such situate nen its calleat Ommsent agent. rY¥Y YE Yr Y YP Pye rrr ys el i, A le ol dk Define Audoneme te fF rahiena\ agent __pennave as per build m sbyewiedge, and clees not yay, attention’ +s \ts perce pes (elnavigiing, envyemment) nen agent is said-do jock Gutenom y. Nu_agent ig said to ve ‘Gurdgneeasi is > oy Cis ef its * a DESigners Are vegarived to ane Al System wort) joulld My Ywontedge as well Os abiity te leon. | Quer agents can work sucessfully im yavielyor ervrspment iC gien soteient Hime to aclapk. | it yh 7 i Dy AElExpiain types (properties) “of _€nyiyonm ent A [Accessible vs Tnacessibie Cowvonm ent . oP TF sensot _debets. -@il velerant aspects of +; + Cavironmenk t6-define choice af an action Then Suvitemmenk iS accessible PEC! Robot diving acam woanls to Cupp’ | only needs mage of the con ahead sp EP ail relevant. inf needed dene a chan oF aor, iS ne available, tren Al systems Uses ts intemal steed dada so Jak convent Behe) then evivirtmmenst-i8 Access} bie ~t EQ: Pacbiing Ade peed not only fee ypate bud al 1ng—_Prrmisaion 2_|Delermoisie vs Non Deterministic (Slochaste eovirooment =p If we ert stale of an enrivenment IS a] Completely. determmed by cunenisliate ana ActICn executed by age wwe say env_ik determin chic j —? Examples Vacuum cleaner agemk worbs in deterministic Cpredicdalte) a eowironmen& becauce en patttychange TF the next siote of an ey eannyt be Complete detemmed by Ane curr slate and Action ig exe cured. ba Ageat we 3d4 Pov iS non olelemmistt TRTGx\ diving 1. ON dedeministic ay we p 1 —=|_ Never predict cathe behaviour LTP TE MPSS] SS {ele e] Sl ey eS) 8] 9] Of Of Of Of Of ofa}olo} ofofolo | ey) ; as soit 7 5 Episodic “ys Noo Episodic Coquenval) env ? PT episod(e agent. experience _is_divided. {ato “9 Z “episode and wal ity of action defenas on conevet Sate. oly —b The net episode “does not depend on acon, taren iA previo episode : Eq: Agen’ sporting dekeenve piece om _Orsembty line =p Tin on episodia” _, cuwent decision’ Cou affect future decis [ons £0: Chess playing and car diving A iStatic ve dynamic : “4 aaagagd PTF env con change while agent 18 perfooming, ouhicy ctnen such eny is calied dynamic env — EG: car accelarahng fon bewind will come cose: ashme goes 36 here agent needs © conhnuesl 4 . a Monto” eaviranment 4 Static env on the omer hand are easy to deals ‘Las ne covrinuos monitoring is. weeded EG" Crossword puzzle . —| Playing ches) _usi ima clock is sem\dynamic Cree a agent “pecemance. ehnanges envirooment) __ BS [Disevete vs Comhnuas_cuvirooment : 7 SIF slate oF env Changes at disrere mera, env is discrese | FP EG! Playing eness is under discrete environment ~H TIF slate of env is changing csmboussty env (8 coohines oS Ea: TAX) Ariving 6 9 » yd : je aa: _ a ee 6 el Types oF Agents 2, Oo. . 0. 4 |Sivnple fefles agent Agent Gensel | what ane wortd | [Lis rant now \ . Y ; a Condttion wnat action —_| : dlovren ule, Should Ta be now | : J - —H Peta =) They qe Very simple anctheinwtelligence is Limite at Agen selects an action basecon current peree ply ignoring rest oF percept history => They wwork 1o Rally Accessible environment —p EG: Thear in frook applies brakes jtnen we pave be jnitiat brake : 3 Such agents work on simple iP-then conclitier Athen Me such aS * fF pin correct , ATM "voj)' Lek you vorthd tau mone yl’ —> CONDITION -~ACTION fUle allows agent te Connetl perce pe ko action. TTT TTT PPT PPP RPP Pe peeeogpood 5 ) oe ee a _ 5 Model Based Agent a = —O = SeaSer °o 5 — ° State) : \ SS Oo Wong world: \ BS How world). is Whe Mocs \ 5 ie evolved | - \ > bo whol my) a [Link] cho | (Emronment= J 7 3 [eonarton =\ Tuonat action = T 1a a [{ Aomen core | [Shouta Tape % | L *| Picinalor ab This model B Usedte Handle _parranyaccessi ble | jefe Slorle of” World as sense? do Hot provide eom He WoOr lA 5 ae ~bLin suan cases sensor need te Maintaly) Some’ —bupaating this imtennal Gate needs 24y varemal state info voice Welps te Jake clecision es oF Wfor~ Ol Ge abou how ne vwrld evolves Cenanging ) Rom behind) _ 2) Hous tne agents 9 own “aeticn affecks + agente sums cat steering. eer Rwise > é We weegh wor Bs 1s called HAG Such model is_mod El basedt Ber yndependentty oF "te agent Ceq. Car accelarerting e “ucla 4 — Ss) Goal based agent | (Siate) Ee vonat me (Hes etee wore) = wees 18 Lilee Mow I ( “attuns “ae BJ a { Lorad tne world wil! | be if Taqhe mis | ark YAYYWIAAAGGAAGIIL + t [Goat J—___\Uonak action shoud LE tebe mow r aL _ _ tems! =p Along wit current sote oF Ww fo, agent necdyrsme | Goal Info_voluich tells what is desiveck situation —p EO. At road ’unehod , a driver can: furan leMing! or go Skaight but conection adectslon elepends on Aestinaten of dani — Agent program inctudles info about result of possible aeton {iq erder te chvose Acton -that achieves Wwe goal => proper search and Planing, ofactten are needed bo Rod ouk way fo aonleve —t The deaspn making sysem looks hoo cond tt? vf “what ei happen “F_AvsSo and x6"! & WI that mabe _ me tntert” wherects in mockel nasecl agent focusis | SIS YYYANNYAY YY Hayy 4 | “iM | on ony ie thts siate then ~med acten', ACh Suppose an age 41g braber uk ! Ben ibs edves not + nol cur itn geal A OOS o oD : How she wend | SS as = now “wana mag 8 agrCn> ) eun do? (ona! id will, be ik T do Anis Adcten wW[C_ ouuts _) oe J & SS [roe haps Torn C Gort) ‘ varmis “sbie i\aldlal d [ere BAO Sra TAake? —} Cohvoher —h Goa) alone are not sufficimt to senerate high qualita behavicue AS gaol _enty specify vohatde be achieved __ ob EG: Many acted se quene will fare dover to dtechnerio buf Hs equally imp to Find out orneper, safer, moe sehable ov cheapest acon seqwente thag omer shone set of action is preferred over otper sek OF action even though bain prencle sane goal .Then His * at sala H yes hishss wtiidy =P UtrIy 1s traefore, 6 fUnenFAINas mndips every eericas vortn evficreot oti teome. : = Tin case of CenPiich goal vdelity funert on _Sipec Aho Node off. EC: Fastand safe driving need | Some +radefP 7 - al al al ara | a ol ol al al QERON that wll Rad te news and iMermertre, lb. expedene te povde beter Solution. se 5: Learning Agent — . 4 9 L peebrmne = | 9 Al standard [Sensot ° {PL tence (coal ‘Y ao | 9. feedback | patente _ Pec rmnare e o F flsaning Giement E T [element | Carre oO. eoig | ere Z 4 2 A |code dee ) | 9 Problem genera ter } | 3 ; ] 5 = > ; Cremater—_Y 2 —4| learning agent has 4 elements > i{ Leaming element: y 3 SEVER A seaveh ajgontnm eres a probe 24 inpub and Te GQ gol 1 ern of aa _ QcHen sequence [Come DePinitense- 1 SBireny space « Te is sero siags which ae pessibic NIG some sequence GF Octo) tne agent Aver fs Catied searol) © Hxe st Seareln yet: TA is q series of Getmtwhnat Ageor af Octuo\ty pedtans fCoa best» Ta deters hein 4 gw slates is goal sine Creamed objective slate 4 4 4+ Solutton: FL is final sAate vartn equi propery s | —__| udile _aptrenal_salietigO has lowe ct pert, ces. among. _O)\_ solution kb le be les 2B Be —_—}+— Co af XPlain vwdter Jug problem wht it Search spare Lmmeasure Drive” waster using Tugs 6f ant 2 lire | As Problem Definition _o Pwoolen Ramnulaeres ap Tn water jug, te siate can be cleseribed aa T pai of vow wotegers Cnjij) Where. X= (041,23, 44 fre rme Amount of waten =. 1a ye jug and yeCo) i€ Ane amount of waten ia Ze | jug & [Represemarien of state Torta\ state = Coo) ~~. sci9 Goal state .€2,n)? ---~ SCG) 3.1 Sek of yules CSearch space) Role 4: Fi) = Bibe sg Cry (37, C4439 Rule 2: Fil UAite — . try) = dud “WR excag Role 3: transcerall water From Ze tol jug (84) =aCaiy,o), (Piro AND xiyc=ee) 5 5 5 5 O oO 7 0 oO 0. oO o 5 0 9 J >) a > > + J > =) J > 2 a a J I J > Pole €: Aransfed come water Bom HL bo Zh jug Rulel: Transrer all water from He to 32 jug. Oxy) = 7 (Ox) FR Cx>O AND xtu<=3 ) Pale 5! Tranefer ome water Hom SL “bow jug wt He jug is Ful) (iy) => Cy teen Eg i Peay ee u) anti Se yag is Rll dey) CX =C3-y>, 3) yf CASO ANWR Arg > Role @. Emply Shjug Ce1g) => Cx 0) iF Lyra). Rule q + Transfer 2 water fori Ze to kd jug (0.2) =7C216) - (f 492.8 xe6 a@dqdaddadeg Problem Solution Grate Cinttal) = (0, 6) >Stote(geal.)+(2$6)12,34, SCininal = Co10) Apphy: - Rule A- ¢ we get (693) Rule 3 4 (3)0) Rulet + ©3713) Rute 5 + (42) Rute 1 + Ce2> Rute A + €270) = SCgoal ) addd edddddddddddddddd | t www ~~ Uninformed . Searching Techni que LEP ne additional Rnouledge oF stale fs given tren iI s|_is Uninformed search or blind seavcln ‘<. Here ing only of stAbore sale , Gopeend)- state and Problere Ramvicriben {s oven. 3 : : pThis technique Keey on &xpirJ Ssucermon (ne vi Ick Ai a solution Cgoal state) is lwcated Bsampe «BES OFS Depin frm senth Tterative Deviiha sedroln C=0S) Wort Cosh Raid Bidheehes U earela. Explain Brs 2 DFs wih advantage and drawbac] DEG L He€ Cook node _'$ expanded fret “men al) successors 6 Me yooh Node Ave expanded next neqtweM aod Successor and go oe ‘ “we can alse say that the tee ic _erpandid breacttn ugise (nonzoniany) ’ can be implemented, by hee searcn using APO queue 4 Tf evmatale has b suwesor (Branenmy foe. stnea +1) dept ¢4) dota no. of Nodes gewa eal ort 1 EEUU EEY EEE EYE YEP eee e ye soy C1000 T | Cet ws be a a Advantages _ of BES _ - ¢ ap TP there isa Finite soln, RPS guararices ik ¢ fp tRinere ise shan one Sotudioo Men sal _¢ ~_ |Ahad we Guiver Mm no. oF SHEPS HOLL be Aesurn al Pipe c | BFS dees not % b oO: i ATL js good. inhnen ne-of borrhes ma nade ore lee ¢ | Disadvantages of Bs 5 ‘ f time. complent és 1S OC ot ') ond peee campleni Sey i Ob") 2 —> Tt needs crane memeny space sire a4) Ve nude y ul Of “Nee = far generated need +o be tered. 2 spat |S wasteful usoen all tae gecl Gere, abe at game level. - A 9g a 4 — a = a ~ q ~ = — a _ _ . ag | a a" | 9 - 5 . 7 LES: _ 7 | 5 “An enponds the aleepess nede first 2 rear deeper 7 sz level bill “modes have nos sucesso Cehild) and. nen : back tracks te next shailow node “that has an | : LS UN. | Sue essor. : a = Can_be impiementeol using LIFO queue Catach? - or recursive funeten (Calling «tne j4s ‘child recursively 9 arf b 1s the branching location Gnd Mm_ys : Movs Sept toa cobalt. “no: Prades generated is bn —3+S e dys iS 6 Ch ‘ Ey 5 is O¢bm) : 5 —} Ectension of DFS is Depth Byes (trie. Sedirch > | Were use can‘limi ne sears PY. Consictonrs 7 -bree tail certoim lew only, J 4 z 7 (A) mao JN 7 (8) QQ ) mal 3 J FY (G: m=) - OV ‘SmemG ee. 7 ® OI DMC) (o o=3 TT ? Lotal_ns.s€ nedes = bam 2 Ly =2%5 3 = ¢ - 3 + - 3 — - — —— a a ried —Rivanages —p sf “= CUvyeph path are stored DFS means Less memory. since say “odes lor cy chance | DES may Find Asan. olor arg ai of ae eae ois =: ‘i meu of hme Oe a “aca 4G [Disadvantages of DES = =p PES can be rapped by. making the wrong : = hece and truck dawn” Very lang poh, DFs does nat guaraniee minimal salutiad DLS : Special case of DFS phe em o€ : avee is sslve : ps — DLS aS Ine node at depth CL) are brated as tN are No sure essor. _ Td May lead 46 incomplete Solution if J WE chuose Unikorm Cost Search oyow 4 y > 2 > The best sein -py BFS may not be the 2. east expense. in terms of path. cost: 2 Se__UtSedth Patna with. whith’ nede to a nd ‘be oJ Q. ih é a vd? Xone wit Jleacd eat 1s -expartted Ave 2. 24 TL eray nek look very offeror but 4 gies 7 of wegvlage te the lew port neol@e 2 TL does net sabe cove, 62 numbed of step 2 i 2 \ Om b oh 2 Sokal pod 7 = Th g1vas sptimdl solutioa 22 4 explores Janet 7 ol g CoS partin Pins. . 7 =p itis equivalent 40 BES and ibs partn cunt 7 {sl feam, 2 : Uniform Cosk porta ( Di (Sy STARE 5 COO 2 i > A An\_- 2 7 Ge CoA > 2 | Step 1 step TE Shep TT > SASS SR=5 Sbq=2 7 _ 2 se=s | site | 2 SC =NS5- SC =15, ¥ Goal STSP 2 : os E DDFS 2 Tecate | Deopemng , Pepin © Fest, “search. - (.2DS) Used nice Q SS 16 6 ° best dept Lint - a > Th us Oe Limit Seah vetth iteration 4 = Depo __limt clarke Aron _O Con\4 oot: node ) “then Qoes Lo A Conly Upto fies, Rew Level DES ) “thes | Spec to 9 (only “upto 2 level pes) and so on S ST has bene Fits of boln DESC les. ete need). 2 RFS Cgiwes optima) elution —> Ti exploves a)! “nodes wn cuwent limit (BFS) * \Cdeptn) oefort “qeing te meet limit (le stn) DDFS. Tiss | pew. (A) a be : ( Om d= BI CO HG da |b - [\ “TN : Of) OW_Wo) d-2 | b few Oe Bee Sb oe Ai ao 6 aad | | Het ‘the hades at the betlem evel are generated ranned) eve -thoke née bette Art_gengled tu ce an SS of updo tne nocd s|_ bel e generic d io! Limes se | =? toda no. oF nowss generated One: : R W2,)O,JOJOjOj2 Ppp. wizogj edb jtd-pP atd-ok | — “Time Com ple “Ha = O(k ) Tt is prefered wren “nee is a large sear space Q@nd_cepln of sol ig Mok Bowe. . | a Be I I& JU JU JU [OY JU 1 JO JY JU JE JO JU VY Ju J9 JO jo ' “| rs ene=s fide < FOO D | at a TAY GB Mere the idea iss nun 2 simultaneous 4 => = predecessors must oe enérated wnich vo 3 search ane fem ingiial slodbe and other bachward fam the goal and stopping wen = “2 searches meet tAthe middle: Benefit 2 tis technique is search me complenity given by OC bal 4. od?" ) is betle® Moan compltsisy of TOES given ly We OC Burl, space__cooipleridy is_an issue because at? A Mme 6Ne of Whe Search ee must be kep? in. memory. £0 we spate Complentia hs OC") voile twpEs: has © () Tt has severe \ynitatkons:— aloalalala Not Always Possible (genacrting root fam chill) = Search must be ce=crdinated belween tnt 2. SEQOWES + = = One Search musi Keep All nodes ia memavy NQ Exarnple ‘ SP Ta ches game .frewy goal state ‘clneckmatd iky officult to veach middie sdate , here * | backward Seorclh needs to Conshucta ———_|_pussiblt_maves be reach 1tS Predeaessorr ws ent eal eal ea! CSHC = et Hote Ea timeaks Gr 7 2U a ee een Metaone = owled fF Gea 4 _\ntecmedia eratjons are given Ment it is “called informed _se aol of Heuristic &voveh i j Climb) eave, AY enrelnin —t Thdimed searching makes use, ot househe Fanction in which addtional Rnewled of the, pdblern of is peeriched, _to search algai\lnen . —> tleuristic function is _@ concept tant astociatec 4 Nove votty every Slate in the given Problem R this help us to _mompare any two slates to_achere, Qea\ evened Search 1 i-€ iL ic a. paramelet - heen strock of a geal ' - a by apling a funchon oO tae slate. ‘ . Tuo —ptlewnste incr _jurct cotculafer clusfance of Inock gate flee tne geal clas 21 The value Assaci@ted ustth each , state is calculated eee ee fie pee epee Spee eyo ye pee pepo pepo pe yore pe > s xr oI] “Ths value calculated by heunstic Pinchors are _ ‘gig —_heunste value —* Meonsts _ thas sepsis foot Shows pyere mm} Heunste- furetiéa wil dive“ siudion) sae peuved {Cou nes “eyataye “ue gate ( 6 Pheawih Soln Dow neceangy mpemrntag sol gic hy ne the effi ie aQ0o0o000 “a Heurisie funetin fer 8 puzele peollem cay be like oO A HH ([Link] Ailes ‘tra are 10 proper _pldre work goose! esp 2 [al | hed 6 ls VZ _ ~ | : =| a = L Ve 5/2 |3 ] ' 5s [2-72 |: 4 7 [he uti TF]: 6 | 8 [4 [ 6477] & El 2) cileunistic funoteo fer erfae doe = HC our usa chantes = eppenen: enti. choye ~ [6 “ a oat > (Hill Climbing Fllgpnitern 2 o Hillclimb search alu) nnowes. AowOnd ste o. sl Qoal usmg _heumstre Tatue CH), it Pinca o volo cht “wi tobe ik closest to Tne oO “goal t:e searok “fr better hence value. qo Te ym piy a loop tad Src rec 0 Hye _olireetov OF yerO asin Value _@& +t oa Tt dows neat maintain “a seomla hee _ —3 + temmmnater when tt-reanhes the peak 5+ | wheré no neighbour has highest value 1 3 iL assurnee to m3 neighkouriyg Aedes so T hare reac he | 2 Tne a AL. a | a— > , =, ;. > > ? | a > _ - - - > - - 2st “ — oe — —— — 2- o _Algactiom:— »|Begin 2 J Mamiain 2 lists : Open = Cstart] & Cisse = . J+ TDaiia\y spen confaane reat noche & id Clase is erp KL 2B | conile OPENI=L 1] do: Remo The lef moss fate. 7 from OPEN call it as if: X= Goal, ten relun culress Ct else: a. Generate chiid af 2» x| b: Puls an clase Ick C! SORT the child of 5 as per heuristic Value of eacn joteC oles encling va}ue.) oP elds bbb debs ds bbe oe cl: toe clildlia ot A pest ir eft side to GPEN 2 repeat; pep 2. Cloop) 4 Ewd Helse end 4 END wine ead < Relurn Foil _ - « nD : may _ oe nana > @%*eCGeav) . ae oO SN Yt 5 Stepl | Open CAD * ) 5 cise C1 4 Step2.| oven C_CS:, R32 D3 4} 4 ae ja clase CA) : 5 3 Sep | omen CES FH 82, 3 2 3 nese CAIC) YLO > ©o© 5 Step 4 | Been [O4,FP2, 02) * 5 at cose Cac, 67 ) > ~ envy: 2 & J; e 2 ss @) ? "I & 5 Sip Qoal__achieved a Sropt Vee el a 9 5 EW Wadueteaur “p- av Br a 9 Hill Chiming on of DAg poss “5 oa mi : 0 LL] Sionpie i Clim a eet +t Seletis the fintt potlor sucressar so itis 4 Less _arcurate “ia Selection. 3 : ; ‘ 2-| Steepeat Will Climbing : 9 Lb seats tne best succ0estoc ge {bt eMcrent wl AM Selecttoo 5 eg} PCio) —> J CiCig), c2ti2), Cara), CUC2r}, = 0 Simple Wil\ climb —> C2. - i Skeepest Wilh climb —> C4 DRAWBACKS of Hil C fimbug := 4d Local Maxima - ds A-local Moanima_i¢ a slate ‘That is beter \ y Store Hh 6UGlh theme _, 1c cheaived _salurtea ive” Q\\oba\ maxima A (pyro e | £ Cl) @)_@) Ca) e>will never reach here . now aadd & ~anananm 5 _ _ " 2 \Plateaun - __ a 3 THs a flad arta vehere all successens ane | oa _| i _, On plate: jis. | Ol 5S} O. more tou loca’ comparison. 2. 7 t . Ban 2. \2 431 > 1 Gi) (C2) (Ca) °. 2 oO. 4 | Kidaes >___ +4 ors are ? > ail \ ' 2 3 2 2 2 2 2 DE S- >» 2 » 2 » 2 2 J — 2 H Gimulated Annealing Search Cc A bill climbing alge tart nemer manos a slot is 4 i i as it will stuck up an lect maim’ Q =? Gone: possible soln’ Can _be_, mabe a random 2 ___| ware je selecting a successer tandemly fom 2 ithe set of-succesmr min mate hill Luni bang © of Complete, hut eahemele ne eR se 9 successor iS sevected, vandovn)ly. Q +The ere, there 1s & need 40 cambine hil | him bi i WALE in some tue nett premides _lootin evfter ent onol a __7 | Complete salutow — Smmulaic annealing is sel an algo. — IT e@ Kn j 1 5 heating The mere! or glats Up ta me)tins pomt amaltuen escling them suddenly 1a =| Cure WEA Wave Gnd size Casal) exleale __ — &xample Task of f ng ~ a | | cumre Call i i «buns sutfate. 2 Nes TTR. toe [eb tye boll ware on Es aut 5 9 will Come & vest orb a (ncal marine e 2.) Bub 56 weshable ‘the curfore we can 9 i toounce om local marima _-Nisletge itde 9 __._ |_ global maxima (share wnutt be Qpprepriate’) 9 the ionermecst= _lvop oF simulated annealing 18 | Ambing phere. Insread oF" § 1. same as nil a __pitong the. bev mene, anandern move ty. pte Deck => TF randem move Gpproves "The situation, 1 is_crceptecl else algo —arce pts move wrth > Same probability ps). 5— his velue _a€ B goes of cleeneanmg vemaon, 5 -}-Os_ we peloct Oot pad mame oe Bad wioves! ane Vifely Ae be allawad - a Y katt luk wot at later pom. oO 5 2 > ~ > . , z » » > 2. s vi > A > > Uy rwr~yyryywy: “3 Aeo ton 0a hale me a Algortom Ceg: Meral Urey. proces 5 Tinputr C prov\ern schedute )- 0 bes 0 Local. VOMAANES : 5 Ccunent node CCN) ) Next Wede Gon) 2 5 Seimperature ~Con ter ¢ tT) D 5 29 ; e for (melt) =1 to 00) 5 £ : ° do 0 = 3 Lind Tle) ; iP Tce. =0 reba CN a a else, _ qd fake NN “ livandomly cerected successon oP CN 7 aq cateuighe A E= NN “CN _ 2 if AG >o % then CN=NN : ese J CIW= NN wit proba bi it e heir’ awa et al ad Y while. Ccdesivecl, ~ pelt luce f ache wel) ole mem 5 =| Deratse here a eer 24 2h ploreok, as tong of Tk sevms promising & sioiteher do atema te 2 bromeh 1? threeme te be mave Promising > x oon suse branth —| Here we L on or net emyand all branche: (\ike Bts) bude branches ae .orepandent aynen needed bn some sttuahcns, ican lead te 2 So 2 | Quaid sdudtad —Ps_sreealy seul often go down bbunally a ercploing ‘The node which tok) promrong Od 9 doves e “ —> Busther provlem_is tat the agent has to s allooep vrécerels of oul explored noctes in orcler te! aveid rtpetitiorn § e —b TP > nace: have same heuristic value bud | Ort. 1 oli erent levOly 2 dig beanthes oF 4 a_4we tne comparing _tvem wifin BES is 4 el _quapra priate a Th cof hot always gae eptimal soln § Qs Gptimal Salutron Mad Wave racer pits atm tbiey scares bacliy Per neunietee Runctrd a MplLonly Cansidec tre pet peunstic valued aca slates tuk ntete congcaers the _é |_arstance ale pone led ann nw J 5, 5 @) Densmmatien 6 4,1 O- Need change ' _@ a wee 7 3. Greedy search: 6+) +1 =% (3 coins) ° “[Link] » HS 2 C2 coins) a. . fo. a Note:~ 2. Hill climbing “Searcy will setk only noker ee 3. A Pfeil —Sepancied—nades olathe _Crevady acd will sett tne eniie OPEN: Lisk. eT TTT i, . | SOT Ces Fa = Eind goal waing qucedy search. CLEEPUPEP EY PL Step 1 oO 0 Open (CA2} @) cluse Co ‘Step = fA. open [65,8405] GOD close FE ATT . Ste pS é Open [Ry , 45, D3, TI] © olose LACT __ i) Stepty = close CAC BR] Ly Gu fS HS .P3,12) ch os GlepS ——— ly ® Open CM1,L5, C465, u5, 03,127] Pea close CAs, BE] XG) Crepe : Goal reached So Stop a nn alelalelalelalelalalel al al ol alalalalalalalalaldlotalalolalala Al ee nm Fip)= 243-0 | $l6)=246 =2 eB WAL f(A) = 24554 | F(a)=24gG = }) SOO a g é 5g I io “sh eareh is 1 Starey) Ts csm bas the bes} wack of Uniform: Path Com Search (optimum complese) and. BES Speed) Ate wvaluaton Rineten_is compiled Os fil o,n2—~ | ftny =*4cn9 + hea’ when a Cgrtead) N= slate Unitrm Coad pein 2 & pets Pawn cost Function (actual cosk Bom rect be cure (nd = Heuristic funeinn Cestimated cos Rem xe cunent -& solutron) Fen) = estimated. casi go Chegpert saludion ‘Through n C Root node 40n) ( Cunent node) a Since gt) aves Ane patlo Cost from sari nucte be Noe Wand hen) {8 the eshmated ciat 6? tne. 5 9D 9 oO oO ° ° oO el a Oo ° ) o. ° Oo 3. 9 3 3 2 3 3 2 a a S 2 2 2 3 2 2 thea pest, om nN the goal we net | Pony "= estnaied Cask af « nec Most im p aspect of AF algo ve i$ its search me st y cpiicnal, cand com ples provided, He ne | Overestimal® (0) tye cok Ao veach lo Cunens nod a aa Caltech act Mais [Link] some as Greedy searcts alge except 14 sosopen tial acceding 40 (tinZra_neng AF is optimal Cquice2 pest) i Sneny i 5 s' Qdnigsable “héunshe “j-e—~Cast— 44 estimates CH) is o : les—than of. gual to actual cack (number of Leo 1 Admissable neunshe hen) Gre ophmiste by ha@tuve and “since gen) 18 shortest (efficient) 4 twlied from root te cument node Ss, o£) never exeved the +e cot of a Suton 4 > So _sduton peadal by At is always Optimal, He hens is adynissible heuristic . s a —__-— A* ts Complete as | Qs i neh Partor CA© avoid “Punning out of space) and. Ronn negative cose A. sockated uatthn each tink (a Disadyvamsages | Al TH+ pas 416 Keep all nade cslates) in__ its merong 4 =| S thcanrun Gut st memory bere} tuna wud 4 7 oF Lime a| 2| This provlem is solved using IDA‘ Cxrteratve Deeging A’\Search’) same ce IDS where erates ____ [was dene based oy deplh but neve iteration Sy 1 done fer chiscrete yellve of — fend : __ S| Example: therdtion 4 Rr £m 10 , jperatien 29 a er Ftn) 420 and so on a —r _ — ~ “d — $< = ni - oS aK = - — ~ ) de EDA algorithm - 7 This algorthm is Aevloped +2 sole mem ogy os regis problem faced by At as AK mee | Om to keep qi nodes n-mem |@nd “tls beromes_- is very ‘clifficult 49 manage incase of lome_omd, oF Compiler state pace op The main olf p/w IDA Ghd pe j ahah _| a of trerchen te lene Por oltsogete- value of Pena Ms \ . o——_] = — ZDAF js goed solution Ry enone problems }© o- ‘ep cost IS positive ond unique forcocle | 5 ETE Oboe cantons are consfeel then 1 DAY Fe | ——_-50._shqet of mom ele tt srefees Same en Uuathon CostPath search ie! has to | + Maratain”ai|_nedes in ners MUS H runs oydoe | Oe. 33 This franteor_ts_sobredl_baf sm AF Csienpli fied | al Memory Rounded. ager n mn) > _ ; (A) S af Cet+e) C2 (ate) C241) ; Level A. Penycie = ACCD 7 ulewe{ 9 2 Femcicm — aRcde > Level fen c20 > ABeperG os 7 _ - = : = ai i Bounded A*? i: Ae : expan ling the best€ {u p ess the 81 Ze fy sess an them Conprsse “os asl maped 7 and cot of rend ; A i Ss tals pestetem bY expand ir Pye “ners is Salutoo at4 ’ ) v " ; @ if ofA Shing Wa disk paging = ys << f <« | =| Craplegraphe Mithmatic Reblen _ SEND | iy jos More _ bach yalueo = OF MoWwey : ae ane, A . Solution al _S HE ND MO R £ 1 Cy 0, to Ant Qy:- ~ yd Worse value oF S) mica Q4a4 T= 18 ‘ al (Cy. =A] a {ME Cy a4] SAMAce « Chye) (4 (ol4) Q+i tO=lo — Att +l =| , 4140-4 treme values | Ptt + t =\y — . Z ot ee E+ Co=N_ : — 2 a o c = _ 1 ~ 0 { 28 | E=N rel _postilele a = 2 a 0 Cyd 9 Es+a=n ° C62 N-qre Consecutive ) is of (ES = a Since $= % of 4, E26 wi\\_ Jove vatue bfw 2-% . a s| No Corry will be generated, : a eC Z=S] { Hg r . NtPtq Se : Net Pt = wl : a RAC ert pe R4IGQ=q . (C1=1 | Cas t&Ci=0 , P=4 not possipie) ~ [Rs] } 25| a DtE=lot¥ 8 Tt Fid : \£E=1 Then as = . ‘ ——— ~ ee “4 O-Compleniif Ayscleptn ak — misraen CDA 7 e” hug i 6 : Qo ws 2 _ ae Compare uniformed Gearch techniques oO i OCriber BFS DES DLS ips Boe Os : ze _ 2 Complete | Yes if bis No“. Ne Mea Tr Besi€ Ou: #oFinite, : bisiwfyne ef oO : batnanean a ce RES Oo” Fo a ; ’ ol ime [on bp Bo B t* leg Prat — iri 5 S ‘vetllowes ys 9 solv. he - 9 45] « 2 J) 2Space | ae Eb be 2 cormplerd Fr 2 2 a 1 26phma) | Yes ifsdep Ne No ~ ep o Vey Wostep Cost ave. Oe Casa den Nea} {Aenkicg) teatro) E bedn drect use BES 2 2 2 oa 3 3 2 2 2 2 2 __Chapber hy continued en page 6 Dy Unit) a 5 axa AL Geant nT - = o AT ; : “0 Game Fla ying : 4 > Tn multizagent env -an agent needs to eowsid of other ent Seni 0 s 1 din achieving his geal 7 Suen env are called competttig, env wre gwe ___|rise-lo adversarial search prob and art _aPten eouttend, as GAMES "1rd adversanal earch prob, lao "playes: ‘Cagents) have «of calat chances put. need opp, sult [er iCone player in chess wid etner wt lose necescas|4. Gaming env are -Pully ' Sis ervabie ‘Ldeleonmnisye) werk - 00 on more playens Cagents). - play aliemertvel y- = Prnysica Gamer like soccrem, orievet Ale FE _ Much comples ako rnptemnen [Link]- Compared ta - lms mile ek | Ames like chess, He 1ae doe ede 7 HE Qame is chePmeol as bind 19 gearch Prob win foil: eomponents : — nitial state “ _tinclucles jnitiay beard position and players mo. Successor Fanetton _ include St Be pation sass Mores fr each state S| les of a game re | 5° . 7 iemina\ gare _ ee — FDetermines Bate ohen a game ‘is over 1 ¢ juss o win or draw (Chess Game 5 Aart “Rinetion : 7 = Ottemines numene value quent termine sid 6 a leteomines Gualtty of tenminat sate, o G 7 = : - vel Game’ tree. - ate te WN i 7 =) - Sarit o i : Oo fee [& LEC N x] oa | i-lst EAA J a on 7 8) z - = * om alot | Fab) i jx be x ES o zy Ct [= =| 7 ‘ ee err 1 5 Tenmiga} ws = (BOT ptt}: tebe 5 Baal eteel Feefel erry + (win) tds 8 Coss) = sC preno) O- any 7 _— a i c 2 a J J J > = - 5 - ¥ 2 2 = Explain _MioMax _Algaiiiam {r-two player ; game using any tee __and pec Alphas be4a | pruning. 2 came sree? , ) ala LT. Z, . a 8 _=}-Th_nemmat search po , optima sol oil oe © _ + | Bequenee: of mares teadliny 46 tennimal stare a buiin game searching ss\t: \s Arena We cause here werg mire made by ptayer Wii! be cependev@ a 6) nis Sppenent neve. =| Tin game dice ophmum Sdutior Cwiantag party) ist detennine my ug af e00 Node ,-yrexe vaIMes 4x8 _assigned_as per MINIMA X® age. : : - 7 MIN IMA: al gonna Crlate cr rook viede as max “pottn current board Shep? | Expand te nade down Ao seme depth C prs % DL =| procedure) till we_getterminat state | p level of necles staring font | —_| Poot, nocle 4o 9" last ot ml2i Node and. minimiemy node Altemahvwty- inne value of Yook moce is computed a a | — « Bap S Backup tne vames Gr coc nen-leaP node wip * u ‘ « € « nade backup veiue ic min ve re 4 minim nization. oe ks. child node oO 5 BBL AL moanimizatin neck€ backup weilue is mdi va o “oF tes chilel nodes . Bo ©. Step Freee the leat Corral pode ieee: whose heed o-——| value_deteennimes Ane Na\ue. st Nook ede I i A ‘ va é o Swot ~ o 5 > 5 1 a xe COG tote we BT 22d OA; a Wiens. perth =ABOT Gnd EEYUYYUYEYYYEYoUOUUDEQade toe do not contribute -Pupscte ‘we calms 0 aye! 9 ed T+ will save vnemory ancl _searen Hime cm — | Te more M is betler than’ move wren player es} 2 | never cert move, N iM pla —l Alpha, Cx) tC cleteermmes lover Bowmal Cort jaa | we. feo “Way ete a e i é nes U 1b Mas Jue for _mininicatien Agee, alalala al NLGORITHM 2 = Fravese ‘the search “hee in OFS. “ Bssuming we "stop Play (revel DD Meng is ect Cakinyy &acdion foe each aNd neu Jos waiue < to its parenk noche. - - ¢ Each non leaf node stove “Dn best packeck waite so far. Best lbacheck value Ab Max Node Ue ¢ | of CALL AS & And ak inlo Node we col MH 18 fi “ | | eA ae Pavers e. min node 2 SL Aces ma\'a) pean d. and this 1s cafled fcuk. 4A | Te parent Adcl® (7) Js c ynva__mede. & xc >p C ‘hen all-the unexplored branches of C will be praned. and twis_is_caited B cut ‘Lei > BOO). Rear eae fain Bt Ee “mae #22, Cl | (Ch jets VN eee] Cant: Fett fer] (ed feed . Peas p= os 1b =25 a aus nm TF Parent. node (P) is a max neae L B(x Ten cil the ‘Wren méredVranche, S&C wr be. pouned anc twis \S cattedl & cub. ~* sad “Recibceralthnen ns . 8g Co) BCI Ay AxeST Dp } PY PPR PPP PHD oOHDD HOH. 1s Re 26 sa oo Tek] [e2 |. 3 Vou fer. [an fez] \ea3y : So 400° Ey yy wT All nodes wich dent Cerbinute te 900] De | WAIL be removedt . a a —= : mn. P28 @) EO (BD Bez = aL) L\ Woe eX [ “ Oo A ® (8) of €B my Og 3 2 2 4 —— S wt aadadaadaalaoa ad adavio o ) — Bhan asc — a a ee ae Y Example 4 po Pers te BF i GS 5 - - = DN eat —> Prep > BOB 3 Prine s& N Ad nee F cc) < XA) Trane q, a Eoample 2. _ aa _ 4 MaKe _ c “93 Be | ° wah $e : a. = PSS # Been eR ne ~ pf 3 15 © SSMS GOO ba aee 2 Ft 13443 476 354 2» 5 6 lo c ; c =. [at node s: ae) > PCB? Se pane | ; c = Jat : A&nede Cr Ble) < AA), So prune FET sy “at | | de al} [3] bebe EI Sa SEP] oh DHS see) FORT AL ep a = to) 4 HR O <4 (el Wal | SUE) nn: se iB! Lee: ey F ERC % : Sl ol R) 4 & : S- S)_| wR] a \ Ie | rifbddr} |* He 1 & a) + ie Kah Ele: u (er Eas St He) J, xa PS Ss a yA g +a 12 Ze ] 2 | ¢! = € Ils PED. Ly £ alet oh re ; a A KC Ih | E Rs) HERR KKK BBE KR BKBKAL AAAR ay pup epee yp Hn). P Sy) ry 4 \ Ly | | | i) 1 Ey LO re AE KDE AY fi ed i) LLL aes NO Z| aT 4 iat). \ | iA z= SI mee lass 2 ATT if ‘fal ® 3 } | wh S A | | | agbOryy) Jf 9 Fe AON LILA Bi ge ler ke @ |x a. 7 TRIS als yj ! - — =f IS te + 7 5 J | E t Ic | gs} £2. AU AG: 5 po > / 7 (Onit 4 tonmnoep t i 5 . ~ i 36: \Constaint Satisfahon Problem (CSP) x : oe os Ob Defined Using sSetoP vamabley IWiiNo,N2 fh o dnd a s@¥ oP consnains Cy,Co =. Cys OT teach vanane nas Non ~eMyphy Domain Cp)'o & 2 sPessile valder i-¢ value dsitaned. var Obes iss Each tonshaint Cor restricts some subnet of. 7 domain watues — Re veedcila ¢ ohh rutiem is : lo vamghite ushicle satiety comin zy lutten 40 CSP-r5 coyreveck $2 assigned démair NaWe je” vOnape Seis ey he Cor taiqls (Condi ton <3 Examples of CSP aire Color Map 8 ou sole nm, Sudoku | Wansportation scrcluing | Hardware Cunfiguratray eke . Exarnple Coler map ' Cs A 4 [ y oa 5 5 5 a z J y . > $ as ] x y 7 a 7 as =| Domain value = 2 Red Green, BlueP | Conshgint => 2 adjacenk regten rannet hue some col t =i FC t Cn “| Cy => Allowed olor fer RiFRo _ Ci— CRB) ¢R.e) (Brg) AIaloivin + (6.8) 62.04) (ae) Y ss 8 Co —> Alinwed: coir RoR BRy 9 Cx 3 gCy F CRIB) C44) 8, BY 9 a I Goe pe ible solution o a ?Ri—R & Q Roa >G a 23 B : - 4 oy Ru > KR $ Ps > G : RE R c Ri R al ——+ Lh queen problem ei] d & Sy Types of Constraints _ me Unary eonstmin ~ st Howe Q. cavtreul gre mng@le van able cannab tak “| cectain demain ved s' [Ext Color of region | Ccannob be red. ; So Ri = 6 green Hed ) L 2 — y » 2 2 z 2 b 2.0 Binary Copstealte 2 + Here “GQ pair of vanabies. cannot dake certain 2 domain values 2 Exl:Tn ¢@ Queen pro & queen cannet be iv » sWatgnt ef cliagonal line _ a Ex 2: Ta coter map , 2 Qdiacentregies can net have LY ‘the same color 2 2 2 2 2 » > 2 > > 2 2 3 Higher ordey + Tt {involves Conssain on 2 of move vomnables aE. Column in Crypt _anitnment mak Shey, Aololi ren constrain posberin 1 Canalse be classified as Ac| Absolute Constain Viclation oF sucin cons tas _usill le out potentrg|varid ) Ex: Tn color map peck 2 adjacent States cannot be Colored, same if this is nok fot lowed, peal) em) a Wail] ner have valich solution " ul Prefered, consvaid | a they cay which soln 2 2 2 = g "i EX Lectomes iQ mem: FI sack Tiicing_pableen (Piggies) = Appiyngg searering alge Cie BEs/nes) ona | HOSP wasies time in Searing sali when Consiva inst “naive already been. defined in: advance = forex: In 2 Queen problena, Suppose we pus tne first Dqueens in top row - Depth- fintesearch will “exomme Cenplove) CN) 8° passiloiiitie? of @ seterhee * position 5 _remamsa ‘six queens before Cliscorenny Mt NO gin Sxsts im same Subiee TRecnskain Gre not Courdend then Rr vamables Waving d vatuer -totat no .0f neetes generated ave W! +k d" which bs - a wastage ar oJ Most OP Neocles will Not Give Solution. Se, we need do insea a tery be Fore wone be 3 { so wanjavies- the sulting alae 1s cajed. | loack packing Searcy. expancing node te cheek wineiner avy consteawrt | has been violated, When valuee are been assyqnect ALGORITHM. , 1! Consicer a CSP pratlem 2 «| Apply packivack Search before Assigning a values bo! The vanable ie check consbraiv) 1S Gh violated, © : 3 | Tf back waeking search 13 success ul Mery gangn Inert: [value to rez, vawable ANA _tortnue step? else cal] * weeursive hark hacking procedure, ~ A a loa K racking preceduc stads A verte Ona art he vale Exampce _ SRackracking fc char map { list of ver v F —_ CRee) => 8, { : Sug meCaD Sap ates fed [oe ee 2-8 DSShO RY vate sehmed eg bs AB, ne ED cE - raleC Ces-8 GEST taak “ | Drawback = —> Sometimes tnts alge dyles to Fina soln to yosched) —p_EX* Suppsse jnattne squares chosen @r the! SEQ queens make ub impassite do place Me a eigin queen 159 the lest column bud bewkracking Uoitl ty ait parsiloie, placing fer Tne sOcariin queen Sve 19.1 An We kno The problew 15 _ymssvero —p PO PwARD cHecMING ALGh _avsids Tots problem by looking ayead ty Abeeo cveailiotl Ay wey wee ey eS SS DT TT | | - : ig + NNOWLEDG E AND Kens NIN G = 1Lrt ofinenwtsey Kinswied ge GE Qctron Anak outcomes enanvy protlem solwing Agent We perferm we ria complex eon inonment: Ae knowotedge ba seel agent cam Combine general Kn ouste: isilol = ire, ko i ci e | aspects of state Ana take Qppropyjate Cerio. - PReasening /inPeming is AvFinect af As ccypacity, fer conious\4 raking cence, 6P things anol i apply. \uqie” Fer estabinshing Q@nck veritging ». AS) Facts availaple i Knuwledge bse. Esampie: Sohn jew tne bneKk 69 the worndlouw and broke jk . : Phere we Know ik wefes to Loindao and not” brick , So redsoning allows uy te Undesdand Ind. al Wer Canrect ly - Rpagtedge representation —Ond_tontinsth = -—_—_} oP Knowledge “bated agents ye . => In order do See _Comyple prob in Al: ome needsto’ ee howe large Gmamt 6F Enowstedge and Some mecha rx Re manipulating that bnantedge Ws create salin te » | New problem 7 — Knowledge. réprexeniation dats Spear hew to 7 jpresenlcnuoledae (fats) 19 KB so Tratt con be $5 eesti manipulated >. Tn ofhenwerde, une wayrk to repreents faebs Cut 5 AF ral World) \n same chosen Rerorcmes y 7 Symbols) the can be manipulated, bey pregcar 6 sf 40a Fer. new knowledge. \. —tthe crotral “Cormpenent knowledge based agent ip Knuwbedge -Base. or KR >: =lA-ce 1s a colectyon oP sealpnear and ea hy >: Semence 1 eepaeerodl Ging Knousleolye ve pres eolat > =| Lanquage. 7 =| Coon seoteee repeserts sore 1% Canerian) of 5 “the _werld. > KB agen} perferns 2 Jacks. = 4-TELL pes KB hut sk perceived usnng sense [2 Ask r Asks KB about’ the aoticn ic be 1) pl mented (pele %. Process 5B. Record execute achion al Pseudo cade 6F CR agent TEU (KB, MAKE = PeRcerT sentences Cpecept,b) ) Hagen 40ils KR what 14 percervea ak Mme tb. Action <— ASK CCB, MAKE-peTIoN ~Query ex ) L_slNs Aske 48 carat acter 14 Siemicl_peoon WZ (wd Qaonny s dene nore by KB 40 sorecLarnadktonD wlalatbatatatatabala lal. TELL (KB , MAKE ACTION ~ Sénrence gachen +4) /] e® achon Selectect by bR Then 4R_ Ag erat vererdo& a \ jZexecule aden and inferm QB of athen pectormak : Let SPTEWMana Ask ye unjien csmy knewled? - al YE presermahen Aanguage. ‘ es chosing cra Spac Kw halt — See + KR agent execute TELL Uony pmpesihenal lagtcé Loe FOL-C Ficsk 6rder Logic Di peekicion logic =FkR agent trecute ASK Yang nosokien , onward *

You might also like