Section -A © Pavanteges of Antifrcal intelligence ) Reduction in human eiart : with Ant seTah ir telgene pthe deetsTonS ane taken from the previov sly Yathened in,oamator applytng a centatn Ss algontthms ° Exam bie — Tn weather forecasting (i) Take aTS6KS Instead ek Humang : This is one of the bigges+ advantage Ast retal intelligence -we can overcome many aTtskKy limitete ns WUman by developin an AL Robot yusnteh TA funn can do siTsky Herings for us. Wi) AVALIABLE 24 %7 ~ using AZ We CaN make Mathtnes Wonk 24 X7 without any bneak$ ancl +hey doniq— even ger baned ) unitkKe humang. WY) FASTER DECISIONS ¢ Ustng AL alongstd® other tehnslogies WE can make machines "take cectston fasten 44, @ human and Comm Out Obons qyurtelcan. Wv) NEW INVENTIONS = AL ie fpowenin many Tnventins in almost eveny demata which © uli help humans gowe the majonity of complex pte blems. DISADVANTAYES OF ARTIFICIAL INTELLIGENCE G) HIGH Costs of CREATION » Machines Need mepanty and maintenance whICh needs plenty ne Co Sts. sey! cmeation smequmes huge Costs as “hey Yasve wane complex machTnes.[Prue =z UD UNEMPLOYMENT? As AL is steplactng the mes 0a tne stepetttve 406ks ancl on cost Ut sebots , hUman interference TS betoning lens Which caTl| Cape A Majost pstoblem 7m e. employment Stand oxtAs- (i) MAKING HUMANS LAZY = AZ. 1s matking humans la With TtHS applicaHons auvtemating +4) maf at he. work 4 ~ only iv) NO. Eto TIONS » Mauhines cannot develop a bone wtth nhumank WhTtch BE an exsentcd attTbute | when comes to +eann Meno emernrt- (v) LACKING bUT oF BOX THINKING: Machines can pen fos1m only those tests which they ee alestq ned ol pstegtia mmed +o do, anythTa out ef +that -they tend +o crash at give itmeleyvan+ outputs wthTch could be a = majast backdmop -PAGE — SB @) IMAYE REW4NITION: one of the Most common USes be machine leaxnnin TS tTmage. secognitton fon dtaital Tmages, the” climensions deseithe the mesutt of “each pixel in +ne Tmage | =n -+he case OF black and untrte im A, the Wntenstty & each pixel Senves a4 a heasune. So TE the black and whitte mage lortstrs N*N_ prxels > then total Wumben prxe and hence- +he ottmenston Te” riz. an -+he Celan image : each ptxel iS Gnsxdened fo provide 3 mea, ement fon the in+ensttHes ob the Bsmam Cebst Omponents, Te: R4KR So -he N* NJ Dian image has S3N2 dimension. SPEECH RECOGNITION = IN sheech metoq nT-+Hon +h e- sof roane- app mecognTrees cpords wands ~The aimenston in HS abplica HON om be asef wwe - Numbens thet me pst1e.sent— the- a can divide the siqnot ™ to poslt+s tenststing ots Arh Renen+- wand ast phonemes .N each Section, we can -mebnesent+ the speech so oe the. indenst+y x1 fpouwen of the 4afgowm greqency tands - hough the debris the siqnal s1epmepenbton = dtede the Bott & tars pow gziasn ) We can mepreselrt the signok shsi0ugh @ set meal values.{PAGE -Y @ @ nas +he fetlowsng basic. data -typ i) Numesttc + Dectmal values ane calied numerics in R- Be fe she Aefou Hr data +ype nUM bone m R you oosTgn a dectmat Jalusg +O a Vestohkie x fellows 7% wll be ef, numenic type - | x= s6 OvT Batnt (ciass¢x)) ti POT W “i 2 “numesic Ratnt (ty be ef) 1 “goubie ” Gi) tureq ER = R Supposrt integen dab—type whithaw the set oh OL wmtegew ‘You con creat ar welt as tenven + A” Vole In +0) Gn TH4ege, aype ustng the as- integen 0) Function: You can eho use the capttal® ‘DL notaton aso Sub pyrx +o denote -+thoxy 2 pantwlian voluzTt a the Th+regen dota type - ti) LOWICAL = R has logrcal data types that take. etme 2 valve of ttve an false- A (ogTea | Votes 1G ok ten cieated Via tom pan ton between vatiableg - P&S whith ane Set- all the compiex numbens-tH e Complex dag +4 be TS + stene numbers wt, on Tmag many Som pon er W) CHARACTER: R Support chanacten dabtype where Yor hove ot the alphateto and spectal enanactenS - Et Stamnes ehanacien values an AAING s- Stim ™M KR can conta Alphas, nUM bens, And symbols . GV) COMPLEX = R Suppasit compie x data +7!PAGE -5 @. The nelatora! openatons ane often used fo Creat: a +eSt expmesston thet con-+t0ls progeiam flow HTS +ype OF exps1ecston Ts abo known ab a Boolean EXpnesstoN bteause they :credtr a Boolean anmswen an value when cvatluated . Relatonak openetoms ane Used -to compare Valves he “Hw0 expneSsTons + RelatHonot opena-tans Qne binary cperertasis because they neque two Spenands te operate An expness@n whTch @nting +he melaHonat opereton TS Catladt —elatonal €x ps1 ten + | Viste a Relational ebengtorts i) 2+ 1858) than ti)c=+ 128 than an equal +o i) >+ qneaten than Ww) >=: qresrten sthan an eqval +o v) ==: equal +o _ vi) f= + NOt equal to:PAGE —6 OTrene anc four ayper of axtigretal wrtetupen ce : meacHve machtnes 7 tratjeal_ memasy , mind and Self — awareness . md Heong 1. REACTIVE MACHINES = The most basTety pes & AL Systems ane bunely react ve, and have the abtiity. netthen -to fO9M memanteg nan +o use bast experiens to Inform = Custmen+ dectsToNS Deep Bie, IBM’, cmess —blayTng Supericom puten, which beat Totenngtonel lanclmasten asnY KaspanoV Tn the. lete 1910! ,% the penfect example of +h ty pe ah machTne . 2 UMITED MEMORY : THIS type Th dass contatms maine can \o0K in +e +he pas+ Self-—dartuing Canis SO Same AthTiS alne - fan exAM bie they obser other cans speed and eons That can'+"be done tn a USF ONE Moment, but athen s1equcien iclen Hy ing specTic abject§ and mont tang them oven yeome. 3-THEORY OF MIND: een ht Stop heney and arth oTA+ the = TMbpandan+ WTAE. bETUEEN the machines ee. have and +he Machines we stl bug 7, the. une’ Howeven) T+ rr betten 42 be rose specific +o ATScuss type S of stepres en BtHons machrnes Need +o keam , and eohat they needto be abeut ~ 4. SELF AWARENESS: The final Step of Ar dlevelopment| ig +o butld systems thet can farm 7ebneLeh-taHons About themselves -UHMimately, we AL meSeanchens wifl have -to not Only undenstond Cons Crouanons, be brild machtnes +hat have 7+ .(senow— 8) PGE = 7 @- surervisen wARNINY — Subenvised lewintng % the. prrcars oly making an algostt-+thm +o jleann +o map an Input +o & partticulan ovtbyt. TATS TS achjeved vstng the labelled clatasets that yeu hows wohected - she na p pT! T commect, the aANTHHM had sucesspully anned: Else yeu can MAKE the } necessany changes +0 aig ontthm s TT Gan jeann Carin % Gampie ~ —rex+ Class iftcaHon bateblems-In +his set ol problems ;+he Goal T% +e pmeditet the clas \abel of oO qven prea & text. UNSubenvTced LEARNING — When We Nass unclasar ied and Unlabeled data the system attempts te uncoven pattenns ftom the data -Then€® 7 no | lowel an +ronget Given fon the examples : | One Common Fast Te to Proup StmTlan exombleg togethen callacl clustentay- eS MEN WU: Retnloncement leannt neler +o qo: = antented on omTthos ; whieh, \eann how © to 44+taTn A complex objective (goal) on maxtmize alen ae pantHwlan dimengir oven many Ssieps- TRIS Method allows machines and Soptwane ents -b avtomeadcally delenniina tne deat ponawtdean setnTe a speetfic “Context TA ander te maxtmte § PoYonrmoants - fan example: Maxinize the pota tS won th a qm pve = omany moves.Pry @: Aq any © Q data Stnuctune that can holel mult -deménsTonal data In QQ, the ro Objects that can hold two o1 mane lan wo — TmenStonal ata: fan example y Ta Savane mati contatn two sews cing -us columns and can Armenston Can teke frve- Arrays On Stow Voluy having only a Stmilan Kind a datatypes. Vnt — Dimenstonal Asia pwhich Ts A vectan 1S a UAT —dimenstonal an Specthted bY a stngle atmenston | ngth- A Vectan con be meated Using (eo) -eneton « Rw let values TS passed +e +he CC) fen tien +o cMeati a Vecton. Examp le : vec} £~c 01,2, 3, 4,5, 6,7, 8,9) Paint (veca) MuULTr DIMENSIONAL AREAY A two —dtimensional matrix TS an asmay Shact teal o fixed numben ef s70WS and colvans j each ontBan TNE the Same data type - Pt mathTx Ts Geated usc ast, C) fuencHon -fo which the valves and e dimensfons wie Prnrcer- Example : arn = anna (2513, dtm = £C2,3,2>) Print Com) -_ (eave ~5_ ® Expert Suctem 1S an TatenacHVe and sneliable com puten~ based dectston ma ing syStem which uses both tS and heuristics” +0 soe complex decision mak tn Rnoblems + Ts Consicdlened at the Wighest level a human intelltgence and expentse- CONVENTIONAL SYSTEM Vs ExPeRT system. ConventHonal system: eknOwledge and procestag ONE combined Tn one urii+- +The Progmamme does not make exsnans “The System 6 Operated only When My Age | . ar oese Batt Thpanmarhon buy ohes Expen+ System : + Knowledge database and the pascenamg mechantam One two" SEbanate Gomponent+s. Pp The expert System may make a niStlce. * Gewtion jg done logr and neusTsHeatty | * J+ can be ncHonal wrth Subkicten+ an mSeotrctert inhormaton : APPLICATIONS of ExpeRr sysSTems. . maHon Management shops} anol Hedtcal factlittes | * Help desk Management * lean analysts *VTNUS detecHon + Warehouse obHmmeqton . Plann tng and schedultiy 2 Steer Manket- trod -LFAWe ~ #10 W) Semr- Supenvtsed learning TS +he sty be macine. lean nTag hak uses &” Combing ton a small ommournt labeled data and w lange amount of Unlad. data -te -matTn models -ThTs a pps1each, +o machtTne. leanntng is @ tombtnatton | Svpetved machine |dannTog ; which Uses Jabelby matning data) and vensubenvised leanntng pwth7ch uses unlabeled ctratatng Aata. A common example of an abpitcation a Senil-supenvised leannt ig om +ext document claxatpten:ThTs Ts the -type of sttvetion Wwhene Sent — Su penvised leannta je ideal because it World be Neanly Impossrble te frnd ao lange amevnt of fabelod ext documents -This TS STmply because Ft TS no+ time ebfreten+ to have. a person stead -thagugh entine “text dowments Just -to assrgn f+ @ Sumple class tRtcaHorn- oo) sent —Supenvised Jeanntng allows fan the algonttnm +o \€ann fom a Small amount sH | 1 elasstp 17 labeled +ext- douments whtle S 4 © lange amount o& unlabeled text documents in the satin ing data -