Scribd
Upload
39 views24 pages

Artificial Intelligence Part1

A.I notes

Uploaded by

rishabhmishra1705
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
39 views24 pages

Artificial Intelligence Part1

A.I notes

Uploaded by

rishabhmishra1705
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
You are on page 1/ 24
madbale ly eke pee AL refers ty simulation of bomen intellyence in machines that ane propummed to think ike humans & mimic thet actiong. tis ord Tyre far PRC aeithane Agron Auto mebile l as Weak / Strang Sepriinkligence a Sheets France Narrow ft Components of Az’ oe or A> + Applications -— Inaye Recesnition + Typer_of Models— Spach Recognition Dasp leering oe * ice Machine lessining Natural angry eloenenton Neutral Networks Sentiment Analysis + Sly 2HIw fon -+naining & running models? — GPUs | panaltel processing tocls( tke spank) » cloud data stonag ¢ compute pletfinms . Programming languages— fython, TensonFlow , “Java , ¢ DABent” approaches of AT | Reichive Machine => These Machines are most basic fam af AT agplatiy §- Deop Blue, Tea's chess -playing Supitompattr. AT teams do vot use any raining sets to eed machines non do eter she date An fine refinences- Based on move made by opponent, the machine decides ret wove 2 wit Meany — pes machine belong tm clase I cepgony of Ar applications. he dniwn cans oe profect example: These ine pr, ie Sore cdlane trained with — othey spead Q d aff (4 nection, lane markings, vane lights une d i isa sr and other important factons, ow time. 3 Titan oP Mind: — Ws is whine ‘we anv, shagging 4o make Gate This concept work hwenor , we are not Jer TOM is fee whee bok will be able 4 “undenstond hormon — emotions, i 2 how they noact +0 them, of AT - powerred caine ano ver to mingle with us & move anound with us, -yndowtanding human behavior is rmpaative- And then, pooacking to such behavion accondliraly is the nopuinement. 4. GR Nisenerte: =" TS ecension of clase Wyre AT Hs one slp aheed of understanding human —ownestons: This is phase whars bubling self aware machines seom n-fhed from whe we shri bday G- Wht — Someone's hon ei From — behind y the machines Should be able 4p feel the emote Tret'c when they Undortiond — ow it fobs pphory tay honk at somecve brck, olvi chess, TSP, TOH infos Problem solving. ea ’ iw paottenn fe IL a PPPOPP POOH OOH O DE MIDIAAN RR AKAMaes FSbbSSSSSISSSSIVILLUVITITUTITT Tyre om ee Vind Bd Ley Grow fst Ganch 5 “Bidineedonal: Soanch Tine |teonti [7 Bat Finck. Seunch [7 A search [7A0* Al gonrthnn 77 thoblem Reduction be Hil) C Nonbing Uninfonmed |B fied Sench!- Tk door contain ony. oyna in Krowledg e Such ag” closeness, location of goal Tk ‘oponetes im — Rnute indlodes infos about how to tra identify les f + goal nod es. Tnfinmed. Sotneh + Tt uses inf seanch . 4 heuntshe Sq wa levels fonee way, as ¢t only. vense the tne £ tow t : demain kraedy The problem is qvatlable whieh om ise the which might net ne be guarenteea fin best soln byt gunteed bh find eek Sh (4p5)-— hy Tine Compl ey b FIFo Queve < may ‘buamchiog. ds dey Space *- 0(44) 4 good Sol ts neaseneble time, ba! tot Vagaeensd i) L my SBG on Sceq © pe py) £2 a | Ae Complede f NaCin finite hep) 4 Th Ophbal Paya ake cement (CF) : Ss a A) 0054-0050) i 3 Complede No Optimal 2 Ay, lonet ae {rng heed) adv, Comp | tle Q Viene ? Spare ? Uptimal 2 eee ele — — as “Donate. Deepening DEG: — B 1s q comptnation of DESBes . TTS aye porfonms depth-finet seach up to a contin “depth limit" & it keeps invneasing the dep Himit aft each jdenation until the get! node it fund « Adv-> This aloo. combiner the denefte of a ors fast search & pects memory Pitieney eperts all wonk 4, ae phase, wel 0 TW B , PL-l dia A.B, c ‘, Mit ME AR e.h a | 0) fl Nevel 2 DLS Ty AB, ry TE, fy Koy ! AO evel 3 ai | m2 : | Rommiches — T+ door Fanwand ela backward . Seancths TH uses BES, DES DLS ck» > One cnt fran ft oe | Adv. 4 fast, Jess memony- and other fen) node» | Dis’ Implementehion is oP» 7 Stop when intensect« 7 Ome should know the gor! ste in advance, . Fostmed@ —— Bavand Backward yee ; , 2, Ops Seren Seaneh Ong “y) @ - ~ ® Sw Howat SN 6 : Oo . Wode Gros node “Tet combinghon of DFEABEC. = Tt uses — hevnishe fanchis hind Tt & ‘mplem ented by prionity gpl SES Aly. => ; tet dee Cit 1. Place Stevrting node nto open [et oY ~ z Ff open it tg empty chp &nekoe filure. 3 Rene node Front 9 im cased li tape? node n 5. Check each jer! nade on Ba ene Anyt pen hit whith has lowest: value of he») £ phon 4 genenrte — sucesons of mde n, Success of node a aad find whether node cp Sueesen mde dy pol wide neta fh, Sveresson node fad Pony 2 @ nade har net bon 14 batty We, adel pote U Retu ta Ste 2) A 2 B 4 ¢ 1 D : ae F 2 H 4 Ts jor llt af: Twitithetion + = Oferta, 6] close [5] - meter pina) clesel 3,6) & NULELAD ctuel'sa7> Spenle,a] Chase PS, Gata es ls AJ CoseP SQ] "3 Open 2,6,6,0) chee 5,8, £] > Sow Et, sn) Cee C8, 84] fi Ba RB SBAAAARHHHAHHDHDONDYWAADARAAK EKA AB AAA > Tt rS mnost Sornmonly known fonnn AC ieeeeragpent— Pes sh , yon) > 1 Tt uses h(n) & Cost to negeh the mde n a t Finck Search 5 ch combined Purtuner of UGS 1 greely be Jit ho) st af path Gt fonerch From n to geal rede n tow Skint shyte Stp-D Place — ctathin ned? i Open litt. Step -B) Cheek if open het ig erapty. Sp-Q Geet node from open hist which Ray Srmellest value ¢ evalvation fanchan Cath ) a Step-8 fxpand a) and genuate tet on into Closed ish. See-@)P vede nt alnedy in open 4 it Should be attnehed 4p back pointer lowest GO) vahoe. Step 0 Return to step-O aN of its succonrang & Closed tse, thery which . reflec ; ) J \ Oy see Inehal 15,83 ; Fewtiont- J (55 ay Ay soos 16} Th: [lssage 1D, 52 A9¢,4) 696 1033 Jy (Srnae Dal) (SONG 6), (55,10) ISahagayy 990496, TE ie used 40 solve AND- dR graphs - is basically based on problem — decom potitiaw - [tiew topes an frame] Reduction J T wsthe | exam | And | as Loy ; Ki Qe (A-aed) (Cah) (0-H) i aod 6 Polenta F(n\zt2 Fon) = 3 AA Climbing Algonithin > i vs a Jota! soarch abn. - Which Continuously move in dinechon of imeteasiing valve to find peak af “mountayy or TTTTINIAAA SAS RR, best solution tothe problem, e- 1 benaigte & Test variant ~ e- = needy approach e- 7 Ne backtracking e. at : e-- oan ; e- A earn ee ee é Stake space Ce a aq "sere sreeevreweeereeeeeoweneeeeememUcmelmUcelUelUce it Hi Chieabiing 45 most used when 4 good hewush’c 4 available. ino” wl Orjece con on loot masini ae a Fist fora) maxinnee? coo tet State Types of Hill Ciienbing’ > TH only evaluates the neighbort Gh) Simple Hil Climbing? = vale sinte at ating & Selech, the ur Sst one which optimizes cumrent ust 2 set it as 4 awwent she )_ Sheepest- Ascent Hill Climbing! This algorithm examines all the nehboving modes 4 the cument State & selects one neighbor node which ix closest fo the goal ste. This ahovithm wnsumes mone time it searches San multiple neiphbuns.- (ii) _Stochastre hill climmbings— does mst etemine Ln i Aces a) j @ ik neighbor behne moving , Rath, this Setnch aly 0 stleck = me neighbor node af vandom 2 deciles whither fo dese it as a cual stofe O71 okomine anther stale, 5 BiB aint ntheton Rexteifeen) > CB oaint Sebi wteny “WA CSP consists of 3 components Nn OV ie set of variables Ta wad re De set o> > C ik set Dorma ins 2d, d, Dn} one fa each vani'g ble. of a that spect fy qllowable tom bination Le of values = (sope, nel) 4 . ;> to) d-] | 2 NE ae = cnypbsitheh Gat), ate] > Two A las), a3 1h) COiRI CRB) CG 18) .. + Two rao FOUR 2 CBR) 3 —> %-(8) (1) - —s Minis Nok Agnes :- - © Hy Tt is a meaunsive on backtriadcing algonithm which is used in = “game theony and deciston -making . > Tt is mostly used fin game plying in AT- Such a Chess, tie-tac-toe, go checkers et. 33 iam, —7 There one 2 phgers Max & MIN. = Max fon mitximized valued Min fn Ininienized value . > miniman purforros a DFS algonithea Poopentieg~of Mini- Mag ?— Limitation !- Complete - es ' Th is Slow fin comple inal’ = Yes games Such ay chess 9° th. Ti tery 0(s”) (oz these have branching Fach So this limitation of mnininngx can be improved — from Space Complexity - 6( bm) eo oe eS Tl] Maninjzor = — 00 Minimizen = +00 ia Termina) values hi Re a, 49 Mae(-l 2) <4 = mor(-)4)” AL rede E C6} Mae (4-9) = 6 = mar(2,6) At 7 a ES-5} mg iw)eng . man(s-5) = 20,74 Marlon 9) 29 * max (0,7) a > 3 > ma xinicey : = Nede B > a minimieen minted. Le Nee Cy 7 maximizer min (-3,1) 2-3 A e Marine 4 Node A: Wi) mart 4-3) oA ? > Th is a modttied version of eninimax algonittm Fe i an optimization dechniue fin mininas 4 Knit. 7 Thee isa Hechnigue by which without checking ath node of Grme tree we can compe Gonnect minima dectsiay and tris th. js called ing. This involves) 2 thnesheld posameten — hg & bets in’ fut exponen. so it Glled —Aipha=sBelyopnaning . Raa sB BRR RB KR HR KAKAaKRKRKRKKZKK KAKA AAA B® H & —. _ Alpha :- The best (hi hest value) choice we hove foond so fur at any point i athe path moximizen. (on Aare) The nial value of alpha jg — 0 Bete- The best ( Jowest valve) 4 minionize- + -f00, (Min phyey) — Ho nemows all the nedec which ane not » -Teally * affcting the final decision « @ — wa yr cH y el win © Alen HE AN 46 Aw — Tominl onde at -y while Backtnacking , mode valves will be peel upper vodes irstead of values of alphe 4 beta. — We will only pass alpha, bet. values to child nodes max(2,3)=3 ,s0 node D value will alo 3+ a= 3 do-# Pe min( 3, we) =3 BIA ; backsnscls Hae max(~e0 3) 25 A=3 C40 F mat (0,3) >5 mar (01,3)23 Foe <—tomin| rode 7 While Backtrtacking, pede valves wail} i alpha - Beja ee ane ua a toed — maK —smin Alpha @s@t G am Re ith re th xe ea ) 7M prsiny pointe WATER Ju PRogiem y— 7 ae ‘Two Jug of diforeet caprcittes one given : in} fe] No maniry there on any Tne gy, 17. #0) 35. S3,3> 40,2> <2. op PPP PPPP PPP HOM HORA ENN A RAK KERRRERe wwuevvwvwwew ew - - Fr PSHHHHSHHHHOHVVOOVYNHHBYUVUEYVY be Pal > Cha) a 2 Ungy= (4,3) 54.1Q) 3 (9) = (4,3) X70 Stele space 2 - ae ‘ 4 Oy) = G44), gro 5+ (m9) = (079) ON (ye) : WUD! (4, Y-(-a)) dro sney Bay (x-G9), 3) | x¥0 A4yy=3 % (ny) 2 Cxty, °) ver(agl = (0 yxty) XFYZ=4 Aiyess (ain) Tile Problem | N- Puzzle Pro bleny > aN plzzle the consicts of Ni tiles Casi) where Nt 1 Na end ei 81S ,2Y andsom Tritisl stae Goal state fists To 4 [¢ |] => #5 Te 1 Eis 5 3 7 bnpty Spare Can move in oP doo ; abt 2 Jeff. (ost ball g eh Using en Stark node ¢ georbes, Pos N58 = \y hea > phea fa © kt la |s\f-3 we al §-purzle | oblem tin Es en rhs, fos - His alg ae3iher tes] oe racer rye Ltrs) tet Goyl node ~ In a chess geme problem, the Start The Gnel on Path a initia) “confiretion of chessboard. ‘) winning any boand configunation , whith & Position ton ony player (there may be positions and each board conhipunstion Can be multiple — final Hest 4 as epresentation q state of the en ever i rie nue Pryor -mever any piece it leads to differed ame.) tt ij . : 'S Stimaked thet chess awe has mone than joe ‘possible “aghites. The gore phying would mean finding * sepuanee valid ynoves whi i a ich bring the bound g shle fp amy of the. possible a we. Pm ma eOSOOHHSHHHHHOSHHSHOHHHRAAKRHEHE? knowledge ase “Ke presealation # _hetening => It i these agonts whe have, Cypabilty af on aiernining a inkl ciate oh Knowledge , reason ove that knowledge , update hein know ledge afte obcenvetions and take actions. = knowledge Base fc megan ed fon updating knowledge fort an qgent ay komm — with — exputevices and take acho as pet the knouledye- Typ: , ? ‘d yt, facts Cy Ache uf bee Declau Ache Crow abot Some ting = a) : Zo Krowledpe ) ye saddd ddd ds dee / ! : Heunrsh'c — Role of! knowkds ¢ 4 Procedural Prot eey meg | é Meta: Nf ernie Ruy dite Snnuledye ‘ ata K ghout k Cha to do Suet) Fe eee yee sss ssdsdddddd (i) Simple Relahons) knowledge (i) Tnheritble knowledge [ivi] Taferential knowledge liv) Procedure) knowledge TH te the Simplest fonm of logic whew — all the stakements aw made by propositions. A proposition) [sq declarative stetimerct which 5 ether dnve on falge. Tt ig a Hechvigae of knowledge nueprestntation yn legical 4d voqthemah'ea] fonw. G- TF is nde OOOO OG Be ws se - _ (A) omic Propositions ~ T+ consishs of single proposition symbol. g. (8) Ceropoond — Rropesitions~ His censtnuckd by carobining simple on qomic propositions wing hyial ‘onnectira @ int ¢ 0) Negation ~P (1) Leplication P—>8 : (8) Grjuncion PAG fqn) —-W) Bieenditions! Pose a Lili) disjuochon PVG cor) e tical retin, age e pene Panes Q AVE & Ase = ea Neetiog wv lomn o AND A _ 2 oR Vv oe Tnplision > a“ ue Bicndition) > a“ PRESSES SSS SS ee ee ee rer ere ere on lorovn dali wily PNG Aesocialivily t~ (pag) ape GAP of (yve)vr- ev (G ‘Vdenhity PA Tes P ' pug = OVP pA( nr) VR) PV Tnve= True Di Shubutive tal avr) = | pra) v (PNR) Pv(@nr)= (Pv8) A (PVR) pe Monin’ laws — aUPAB) = oP V8 APVO)= oP nog Double nega Hon 7 alap)= Pp ~PRpoitangl Tyids— We cat nepnesent nelations like ALL) seme 07 none: Y-(q) AN the ginks one (b) Some apples cure FOL ws a pt inkeljent sweat Logid => Th is an extension to propasitiong! leyie. cwerl language tnt develops ieemation about the objec in a mone cay 19 and can alto express the — nelationship kha those sbjeck, nelatans & function « 7 FOL alo has 2 tmin pair f? 8774 > FOL assumes tach, Syntax of FOL oF Constont Neniables > Phedicatet Fundien _, ba nectives tylity Soon by > objecs- Semonhig \\2, A --- Rt Tay b-- — Brother, father ---- syn oo a NN AS Vd a SS Momic Sentences:— These sentences ane formed from a predicate Poh Symbol fillowed by pananthesis Paedicae( t1, by ~-- bn) J REA ane bethow =>" Brthas’ RA) Comples Stolen The ane made by combin’ Sentences using Connechyey, ae Is an intgan Subj ect Predicate Buantifiers in FOL:- oe Poke Ty ahmic (8) Existential ( tn seme, at lent one) Vx man») —> dunk ( x, coffee) Any bystx) eA intelligent (x) Propenticn of Quantfins— — \Yx Vy = Awa Andy - dyad 4 < é C ‘ « « “ < e « (9) Universal (fn all, everyone, everythi « ( von veyything ) v : « « ¢ ¢ € ‘ ‘ avy + Wax — ‘ c « Te about the decicion makin | tasks permed NObofs oh Lompadeyt PApes to achieve 9 epecihic goal : u ! ‘ ( The execution — of planning © about choosing Seuanee actions 4 yh Iikelibood — ta complete the” epeaiie fuk. 2 ; " | Stadt state Geo! Gate | broal Stack planning i - I+ 1s Speen ally used by STRIPS. Stack is used to hold achon & sarify the ul. A knewledye Base % used to hold curneitt sta, achon —streel statk i Similan to a mode in a seanch tree whine jhe bnanches ane caeged if there & a choice ff an action Nor. Linewt planning: — This phaning is ged tr set a goal stack and 13 inceded in seanth space af all possible subgoal ondening s- Mncentainity > Suppose if A is dove then Bis tnve means AB . In this sitvation we can certain. But fwe ane not sure abut whether A is true on net then that sifvahion is called _uncortintty Cs cane af “Uae (i) Tnfrmagtion verumned from unneligble — sownces. (i) Experimenta Svrons (iii) Syuipment fold (i) Tem. variation) (Vv) Climote change. Probabilistic Reasoning = It isa woy of knoledge representation whine we 4ppy concept of probability. to indiate the Uncertainty i4 knowledge - We combine probability theony with lege do handle vncentsi ty wi solve problems: — ° Bayes! Rule + , Bytsien Sta hishieg — eelied mig nteriat, OS PCA) £1 conhinty PlslB) = Plans) P(A) +P(A) =] p(B) KKK FF FSHSSCS SSSI ss ests sd ds ddd ds dddrar re epee eres seeeaeeseeeeteaes fesse rare a, = PCAAB) P(Ale)- Ean , pt Blae 0A) | tis. babilty “fan “Bayes! Theonem > Ti determines the — pro acess event with Uncertain knowledye - nth ST} & a way do coleulate the value of PC Bla) w \ knowledge of ptale) 7 4 ie updating probability prediction of ay event y observing new ——trformation of the neq) wonld. POAAB)= pCAlB) PCR) on P(AAB)= PCBIA) P(A) P(AIB)- PL BIA) Pn) L P(e Tt howe simple —nelahinship blo joint & conditional prbabilities- PEnB) ig poctenien PORT = PCA) Peas) Z Pei) Plein) — >It tha Bayes! Rule. [ Jeyenenl peed = PUBlA) rca) ] | BE whet is the probability that patient has diveayey meningitic with q stiff eck? A doen vs quant thet disease meningitis causes a patient te hae a sti neck 2 i ocuns ayy. time, * PEDM= Ween: 5 pes) = 3 : PC sib). og Pls)= 6:02 | P(Dy= —l 3c0ee PCo|s)= PL sto) pep) | gcgee ics 30000 aa. en on ne PCs) DL 83 Ee ee ll er rlerl el aelhLh eOrellhUl OO re samen Baye Sate Nedoorhn> Th isa probabilistic popbial (decison [belie ) model which represents 4 set of \aniables and their — Conditional dependencies wing q dinected ache aaph> [Dab] Tan be used fon building models from data 4 experts Opinions and it — tonsisk 4 2 park: © Dinected Acyclic Graph * Table of — tonditiona) probabilities J ede [parent of 6) nv © G ent Tt has maine 2 coenponenss: oe ~S Act] Numbers To — propagate be lief in Bayesian Ni initial Dineched Acyclic Graph" is wnvented nto an ondineded graph in which fhe ana can be ysed te transmit —pnobebih in din of evidlence Helg 9 EE ~~ (4) Nena] Nehoonks :- There ane tornputing systems lospined by the bivlogica| neung) — networks that constitdé animal — bnaing. The — idea of ANNs ig based on belief thot il att homen brain $ by hg vught connechong , Gan be imitated | wa Siliwm & utnes as Nhuig yevnon’ 2 —dendwiteg. Hidden sn ‘ » 4 ‘ ) , ) ) ) ) ) ) ) ! Vypes ae ANN: (i) Feed Foruand ANS? A vit fram which jt Tyee onk = ao te N \) Feed Rack ANN: An: addnesca ble memong. We ivfymation flow & ord inedhione] unit sends information fy other does rot neceive any iesporengtion - back — oops. ® Feedfinand Anis! => Feeback loops ane allowed. Thy ene ured in wenden 7 Cbs Seed Sith ANN mn he nheanhkhnhneannnwemeaeeneeeeeeeseeeeeeRemeteeeeeeeeeeeesh,

You might also like