rtificial intelligence (Al) Aertercm Le wrewtoence Artihiel ntetigence is the obitity fo computer or a cobet controlled by ¢ computer Yo do tod thot ae wally done by humans hecoune, Wey Cyive huwon iotetigence Drsroran’s Auasesrues Csingk souwe short goth) wore: Mops , telephone networl .sociel netvorking Ove OM represented by gaophe (noice) and to find the shortest goth between oaks we use Dishstra’s Aboritn 2 v tt dtu)s cluy)< a) Ae © oA* lv) = slays lawn) & TO o+m< ve 2040 Oruo cw B12 No C UO ve osue v2 20410 veo vio[—[townt | Town? Town3 | Town4 | TownS | Towne | 3 5 2 4 3 8 (5) Sra: Tnitilige Bare to 0 oe Nodes to en alicgroor smr2, Make a tuble Bo | te va 8 me 0 a uss 2» © 3 5 2 o 2° 2 c 2 q 32AL’ Aissossroes AA” Alguith is bosed on Dighitio , bul adds on extra heuristic value (Ao etcliged gueis) on hom for we bore to qo to mad the destiodton most efficiently wz w ©) ez Att ct lore) fe = os Meuiiewke < S / {oy Aw) Fed the ghortest goth to rach G from $ ati ewes Peo) = va + ho) ve ‘aetuak ? 2 cost:SB fr g(anlo) se Feds qledy nfo) 2 U4le Bau 2 tb Cd sce seo See SRE Puy. quash Ra)» ge) $62 gedH6) fon gle) hin) BHOr 4 2 hb Sr +il 2 MANDY .o @® . 2 120 seo Hier _g(a)a Wo) 2 3470204 SCDEG en)» gl) Kon) BrTHWUS +O a Mecucne Leaawine Arvo Deer Leaman,telligent machines think and behave) r ti ke humans. ) L {Systems learn without being programmed ) {to learn J) eae think in a way similar to the } ~ s human brain; they handle huge amount: \of data using artificial neural networks How actficiok nemrah networks have helpeat sith machine Neoraig & Neural Networ hs contol Be am Stearns from post epee Aarrercret Neurol Network mmuchnesen_ceubed to tare Rom post opeicnce (date) (Baby 4 ) fovit s aaa Mectine Leasing +1 a subset of AT in whch alyorithns ait Moines ond fear Fram their past axpuiene 4 con ben oy doko thet Nigh Feequzecy ashe al worl Be heLostreed Avo Unasenco Dara lUnatates doko, The doto Co which tag or Oabd ix rot obtachedt ltebeted lake The uslobcttes termes Coboted deta the moment a meaning i otteched (Label 13) Wees Or Macurve Lennie, D_ Supervised Goering ® Unsupervised Larving @D__ Reinforcement Searing (ee Sie Labeled Jebe— UnlebMed Jats Gh Dog fue date seth Rarring Rsihicodion divides in qeore’ by brink end vin Pod bern Supervised LeanSupervised Leann Usieg Wroun bass vaith given oukcomes to enchle a computer program i to improve its performance in axonplshon similey bask Unsupervised Dosey wsiog a Qaege sumber & tudes with unkeawn oulcores to enable la_compettes progam bo improve its peformance in exconp hiding similor bk Reinforcement: Naceieg using a Rory number of tasks with unknown outcomes and the use of T feedbacs to mable a compller program te Improve its pefemarce in saontlishng sini Por tock Dees lenrrane Cis_y sunstr oF macnive Lenrtner) Deep Reoming is 0 type ot machine causing inspiced ‘4 the _shutlwe af humor brain this steuctace is cobed Artiticioll Newel Naworks + Exbrack pattern From doku using neural ackworks }Macume Leaawins Fiotues (Boys COtne)) Cats Cin VY 0) cee Lenentwn ag ae TRS + AtiticinR reveat nckuorks are bered on the interconnection between aearont in the humen broin. The sy ctem is able to think like harmon using thse neu actuok and its, pedocererce improves wsith more detec | The bidder Sayer are where doko from the input Layer is procesied into something which com be sent to the eukput foyeron [| Se weighted Thaw Stoke the ccosen For hovieg multiple hidden Moyers i9 on actifind nevrcl ockwerh” + robles deep Sesccina, bo talae place + Where the problems you are trying fo solve _has_a higher evel of complenity it cequires more Coyers Cocdve + To enable the newrot networts to tearm end oske decisions on ils oun + To improve the acuwracy ot cesults, ‘machine learning deep learning ‘enables machines to make decisions on their own based | enables machines to make decisions using an artificial ‘on past data neural network needs only a small amount of data to carry out the the system needs large amounts of data during the training training stages ‘most ofthe features in the data used need to be ‘deep learning machine learns the Features of the data ‘identified in advance and then manually coded into the | from the data itself and it does not need to be identified system in advance Bac Reoromarow Ano Ressessron EaQainobion Precastocie i Pek iy Peake lWeecastc, Riadicting soles for this yea > ecight: Neurol Network takes Condon weight voluc Peeastoc Redic y bop Hidden Ouke uk 200 isu RPO wo toss 200 \sw use 160 Roa Algorithe reduces Mis escor by pdabing 7 t the_waght -Treisoy_proqrom ts iteabiweTian program cs ituabive |. “Re es a hs bodrrnto the peucod nelworks te order tO update the witicd netwaring oo tings |. This —-— ws. tepeokied watit the deswed outputs orc cecrhuclly obtained + Describe the back pinpegekion p eacess The initio eutpol's from the wysten afc compared to the oxpeded onkpellm The syste seeghbnys se adjusted Xo _minienige the diflegeate between acm ok, and expected result Once the prrocg in the oukpuk hove been pDminced the seared network as fundioni correctly If the error ore stil too Borge the ecighlngs ace_alteed. Recre scion -\s one of the way of 1 dake before irs wnpell mks « system \s wed to mabe geedidions From given dale by Lang sone colhionse between the input and the oubpde