{"id":52119,"date":"2019-11-01T00:00:00","date_gmt":"2019-10-31T21:00:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/blog_prohoster\/kak-ustroen-gibridnyj-igrovoj-ii-i-v-chyom-ego-preimushhestva"},"modified":"2020-02-18T13:59:47","modified_gmt":"2020-02-18T10:59:47","slug":"kak-ustroen-gibridnyj-igrovoj-ii-i-v-chyom-ego-preimushhestva","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/novosti-interneta\/kak-ustroen-gibridnyj-igrovoj-ii-i-v-chyom-ego-preimushhestva","title":{"rendered":"Si funksionon AI i lojrave hibrid dhe cilat jan\u00eb p\u00ebrfitimet e tij.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

N\u00eb vazhdim t\u00eb tem\u00ebs s\u00eb ngritur dikur n\u00eb blogun ton\u00eb t\u00eb inteligjenc\u00ebs artificiale n\u00eb loj\u00ebra<\/a><\/noindex> do t\u00eb flasim p\u00ebr sa e aplikueshme \u00ebsht\u00eb m\u00ebsimi i makineris\u00eb n\u00eb k\u00ebt\u00eb fush\u00eb dhe n\u00eb \u00e7far\u00eb forme. Ekspert p\u00ebr \u00e7\u00ebshtjet e IA n\u00eb Apex Game Tools, Jakob Rasmussen, do t\u00eb ndSharej\u00eb p\u00ebrvoj\u00ebn e tij dhe zgjidhjet e bazuara n\u00eb t\u00eb. ndau<\/a><\/noindex> ekspert i \u00e7\u00ebshtjeve t\u00eb AI n\u00eb Apex Game Tools Jakob Rasmussen.<\/p>\n

\"Si
\n
\nN\u00eb vitet e fundit, ka pasur shum\u00eb biseda rreth asaj se si m\u00ebsimi i makineris\u00eb do t\u00eb ndryshoj\u00eb rr\u00ebnj\u00ebsisht industrin\u00eb e loj\u00ebrave, sepse kjo teknologji tashm\u00eb \u00ebsht\u00eb b\u00ebr\u00eb revolucionare n\u00eb shum\u00eb aplikacione t\u00eb tjera dixhitale. Por nuk duhet harruar se loj\u00ebrat jan\u00eb shum\u00eb m\u00eb komplekse se sa nj\u00eb simulator makine, nj\u00eb program p\u00ebr menaxhimin e dron\u00ebve ose algoritme p\u00ebr njohjen e fytyrave n\u00eb imazhe. <\/a><\/noindex><\/p>\n

Nd\u00ebrsa industria e loj\u00ebrave ende preferon t\u00eb p\u00ebrdor\u00eb metoda tradicionale t\u00eb IA, si\u00e7 jan\u00eb metodat e automatik\u00ebve t\u00eb fundm\u00eb, pem\u00ebt e sjelljes dhe, s\u00eb fundi, gjithnj\u00eb e m\u00eb shpesh, IA e bazuar n\u00eb utilitet (sistemet e bazuara n\u00eb ndihm\u00eb). K\u00ebto IA quhen gjithashtu IA t\u00eb dizajnuar ose sisteme ekspertiz\u00eb. Por po b\u00ebhet gjithnj\u00eb m\u00eb e qart\u00eb\u2014dhe p\u00ebr lojtar\u00ebt n\u00eb radh\u00eb t\u00eb par\u00eb\u2014se k\u00ebto sisteme po b\u00ebhen gjithnj\u00eb e m\u00eb pak t\u00eb p\u00ebrshtatshme p\u00ebr krijimin e kund\u00ebrshtar\u00ebve t\u00eb avancuar q\u00eb mund t\u00eb imitojn\u00eb sjelljen e lojtar\u00ebve. Kjo vlen sidomos p\u00ebr zgjidhjet kreative. Kjo mund t\u00eb shpjegohet me faktin se zhvilluesit e inteligjenc\u00ebs artificiale nuk jan\u00eb n\u00eb gjendje t\u00eb marrin parasysh t\u00eb gjitha taktikat dhe strategjit\u00eb e sjelljes dhe t'i realizojn\u00eb ato me sukses n\u00eb sistemet tradicionale t\u00eb IA. P\u00ebr lojtar\u00ebt, kjo shpesh shfaqet si nj\u00eb p\u00ebrvoj\u00eb e m\u00ebrzitshme dhe e parashikueshme kur luajn\u00eb me nj\u00eb kund\u00ebrshtar t\u00eb cilin e zbulojn\u00eb leht\u00eb.<\/p>\n

Ky rezultat vjen si pasoj\u00eb e shum\u00eb arsyeve, por nj\u00eb nga shkaqet kryesore \u00ebsht\u00eb paaft\u00ebsia e IA p\u00ebr t\u00eb m\u00ebsuar. Prandaj, kur krijohet inteligjenca artificiale p\u00ebr kund\u00ebrshtar\u00ebt, natyrsh\u00ebm aderon zgjidhja e kalimit n\u00eb m\u00ebsimin e makineris\u00eb, i cili ka pasur rezultate t\u00eb shk\u00eblqyera n\u00eb shum\u00eb aplikacione t\u00eb tjera. Por ka disa nuanca q\u00eb duhen marr\u00eb parasysh. K\u00ebshtu, IA e loj\u00ebrave duhet t\u00eb jet\u00eb n\u00eb gjendje t\u00eb adaptohet ndaj \u00e7do situate dhe t\u00eb p\u00ebrdor\u00eb p\u00ebrfitimet q\u00eb ajo i sjell, si dhe t'u p\u00ebrshtatet stileve t\u00eb ndryshme t\u00eb loj\u00ebs s\u00eb kund\u00ebrshtar\u00ebve\u2014lojtar\u00eb t\u00eb gjall\u00eb dhe IA t\u00eb tjera. <\/p>\n

<\/p>\n

Si si ndodhin gj\u00ebrat tani<\/h3>\n

\nKompania britanike DeepMind, e angazhuar n\u00eb zhvillimin e inteligjenc\u00ebs artificiale, s\u00eb fundmi tregoi se si IA mund t\u00eb m\u00ebsoj\u00eb vet\u00eb t\u00eb luaj\u00eb loj\u00ebra, t\u00eb m\u00ebsoj\u00eb rregullat dhe t\u00eb gjej\u00eb m\u00ebnyra p\u00ebr t\u00eb p\u00ebrfunduar loj\u00ebn ose p\u00ebr t\u00eb fituar\u2014eshte e v\u00ebrtet\u00eb, deri tani vet\u00ebm p\u00ebr loj\u00ebra t\u00eb thjeshta, si loj\u00ebrat Atari<\/a><\/noindex> \u2014p\u00ebr shembull, shpella dhe loja logjike japoneze go. Rezultatet e arritura p\u00ebr to tregojn\u00eb se inteligjenca artificiale \u00ebsht\u00eb n\u00eb gjendje t\u00eb formoj\u00eb nj\u00eb vler\u00ebsim t\u00eb qart\u00eb p\u00ebr at\u00eb q\u00eb po ndodh n\u00eb fush\u00ebn e loj\u00ebs. N\u00ebse flasim p\u00ebr adaptimin e IA ndaj stileve t\u00eb ndryshme t\u00eb loj\u00ebs s\u00eb kund\u00ebrshtar\u00ebve, rezultatet deri tani nuk jan\u00eb aq impresive.<\/p>\n

N\u00eb dit\u00ebt e sotme, rrjetet nervore tashm\u00eb kan\u00eb m\u00ebsuar t\u00eb njohin imazhet dhe t\u00eb drejtojn\u00eb makina. Por k\u00ebto funksione mund t\u00eb realizohen edhe me arkitektura relativisht t\u00eb thjeshta, edhe n\u00ebse si rezultat ato do t\u00eb rezultojn\u00eb mjaft t\u00eb thella dhe t\u00eb m\u00ebdha. P\u00ebr shembull, IA p\u00ebr njohjen e imazheve n\u00eb Facebook ka nj\u00eb thell\u00ebsi prej rreth 100 shtresash, duke e b\u00ebr\u00eb at\u00eb t\u00eb ngjashme me trurin biologjik\u2014p\u00ebr numrin dhe kompleksitetin e nd\u00ebrveprimeve midis neuroneve q\u00eb formojn\u00eb nj\u00eb rrjet t\u00eb madh. <\/p>\n

<\/p>\n

IA e loj\u00ebrave<\/h3>\n

\nSa i p\u00ebrket p\u00ebrdorimit t\u00eb m\u00ebsimit t\u00eb makineris\u00eb n\u00eb industrin\u00eb e loj\u00ebrave, ekzistojn\u00eb disa kufizime q\u00eb nuk lejojn\u00eb gjithmon\u00eb p\u00ebrdorimin e k\u00ebtij lloji t\u00eb arkitektur\u00ebs. K\u00ebto p\u00ebrfshijn\u00eb k\u00ebrkesat sistemore, t\u00eb cilat lidhen posa\u00e7\u00ebrisht me CPU-n\u00eb, t\u00eb cilat p\u00ebrcaktojn\u00eb aft\u00ebsin\u00eb e kompjuterit p\u00ebr t\u00eb p\u00ebrpunuar struktura komplekse lojrash dhe p\u00ebrshtatshm\u00ebrin\u00eb e tij p\u00ebr tregimin e loj\u00ebrave dhe gameplay. <\/p>\n

K\u00ebshtu, rezulton se n\u00eb shum\u00eb loj\u00ebra nuk ka mund\u00ebsi p\u00ebr t\u00eb zhvilluar nj\u00eb sistem komplekse t\u00eb inteligjenc\u00ebs artificiale fal\u00eb infrastruktur\u00ebs s\u00eb nevojshme dhe, akoma m\u00eb shum\u00eb, nj\u00eb klaster serverash, i cili ekziston, p\u00ebr shembull, p\u00ebr rrjetet e njohjes s\u00eb imazheve n\u00eb Facebook. Ndonj\u00ebher\u00eb, disa IA gjithashtu duhet t\u00eb punojn\u00eb paralelisht\u2014jo vet\u00ebm n\u00eb kompjuter\u00eb, por edhe n\u00eb pajisje mobile dhe n\u00eb platforma m\u00eb pak produktive. E gjith\u00eb kjo imponon kufizime n\u00eb madh\u00ebsin\u00eb dhe kompleksitetin e arkitektur\u00ebs s\u00eb m\u00ebsimit t\u00eb makineris\u00eb, sepse t\u00eb gjitha llogaritjet gjithashtu duhet t\u00eb realizohen me nj\u00eb afat kohor prej rreth 1 ose 2 milisekondash. Sigurisht, mund t\u00eb p\u00ebrdoren teknologji t\u00eb ndryshme optimizimi dhe shp\u00ebrndarjeje t\u00eb ngarkes\u00ebs midis \u043a\u0430\u0434rove, por ende nuk b\u00ebhet e mundur q\u00eb t\u00eb eliminohen plot\u00ebsisht k\u00ebto kufizime.<\/p>\n

Problemet serioze p\u00ebr AI mund t\u00eb shkaktohet nga kompleksiteti i loj\u00ebs. N\u00eb loj\u00ebra si StarCraft II, mekanika e loj\u00ebs \u00ebsht\u00eb shum\u00eb m\u00eb e nd\u00ebrlikuar se ajo e loj\u00ebrave Atari. Prandaj, nuk duhet pritur q\u00eb, me nj\u00eb frekuenc\u00eb kadrash t\u00eb caktuar dhe me k\u00ebrkesat e njohura t\u00eb sistemit, m\u00ebsimi makin\u00eb t\u00eb p\u00ebrballoj\u00eb t\u00eb studiuar t\u00eb gjitha gjendjen e loj\u00ebs n\u00eb m\u00ebnyr\u00eb t\u00eb plot\u00eb dhe t\u00eb mund t\u00eb nd\u00ebrveproj\u00eb me t\u00eb. Ashtu si nj\u00eb lojtar shpesh drejtohet nga intuicioni n\u00eb fazat e hershme t\u00eb loj\u00ebs, AI gjithashtu duhet t\u00eb m\u00ebsoj\u00eb p\u00ebrpunimin fillestar t\u00eb gjendjes s\u00eb loj\u00ebs p\u00ebr t\u00eb leht\u00ebsuar kalimin e m\u00ebtejsh\u00ebm. P\u00ebr shembull, n\u00eb nj\u00eb nga t\u00eb fundit API p\u00ebr Starcraft II<\/a><\/noindex> n\u00eb hartat shfaqet vet\u00ebm ajo informacion q\u00eb zhvilluesit e konsiderojn\u00eb t\u00eb r\u00ebnd\u00ebsish\u00ebm: n\u00eb nj\u00eb rast AI p\u00ebrdorte nj\u00eb pamje t\u00eb zvog\u00ebluar t\u00eb gjith\u00eb territorit t\u00eb hart\u00ebs, nd\u00ebrsa n\u00eb rastin tjet\u00ebr, si nj\u00eb lojtar, ai mund t\u00eb l\u00ebvizte kamer\u00ebn dhe at\u00ebher\u00eb perceptimi i tij kufizohej n\u00eb informacionin n\u00eb ekran.<\/p>\n

\"Si
\nVizualizimi i loj\u00ebs AI AlphaStar kund\u00ebr lojtarit n\u00eb StarCraft II: n\u00eb ekran shfaqen \"t\u00eb dh\u00ebnat\" fillestare t\u00eb v\u00ebzhgimeve, aktiviteti i rrjetit neural, disa nga veprimet e mundshme dhe koordinatat, si dhe rezultati i parashikuar i ndeshjes.<\/i><\/p>\n

Ky \u00ebsht\u00eb nj\u00eb aspekt ve\u00e7an\u00ebrisht aktual n\u00eb rastin e loj\u00ebrave. Shpesh, metodat e zakonshme t\u00eb zgjidhjes s\u00eb problemeve t\u00eb m\u00ebsimit t\u00eb makinerive nuk jan\u00eb t\u00eb zbatueshme p\u00ebr inteligjenc\u00ebn artificiale t\u00eb loj\u00ebs. P\u00ebr shembull, zakonisht ai nuk \u00ebsht\u00eb i detyruar t\u00eb fitoj\u00eb ose t\u00eb b\u00ebj\u00eb gjith\u00e7ka p\u00ebr t\u00eb siguruar fitoren, si\u00e7 ndodhte n\u00eb rastet e loj\u00ebrave Atari. M\u00eb shpesh, roli i AI \u00ebsht\u00eb t\u00eb b\u00ebj\u00eb kalimin e loj\u00ebs m\u00eb t\u00ebrheq\u00ebs. Ai mund t\u00eb ket\u00eb nevoj\u00eb t\u00eb luaj\u00eb nj\u00eb rol dhe t\u00eb sillet ashtu si\u00e7 e k\u00ebrkon karakteri p\u00ebr t\u00eb cilin \u00ebsht\u00eb p\u00ebrgjegj\u00ebs. N\u00eb k\u00ebt\u00eb m\u00ebnyr\u00eb, AI-t\u00eb e loj\u00ebs jan\u00eb m\u00eb shum\u00eb t\u00eb lidhur me dizajnin e loj\u00ebs dhe tregimin e historis\u00eb dhe duhet t\u00eb ken\u00eb mjetet e nevojshme p\u00ebr t\u00eb menaxhuar sjelljen e tyre p\u00ebr t\u00eb arritur q\u00ebllimin e caktuar. M\u00ebsimi i makinerive n\u00eb form\u00eb t\u00eb past\u00ebr nuk \u00ebsht\u00eb gjithmon\u00eb i p\u00ebrshtatsh\u00ebm p\u00ebr k\u00ebt\u00eb \u2014 prandaj, \u00ebsht\u00eb e nevojshme t\u00eb k\u00ebrkohet nj\u00eb zgjidhje tjet\u00ebr. <\/p>\n

<\/p>\n

Problemet praktike t\u00eb m\u00ebsimit t\u00eb makinerive<\/h3>\n

\nK\u00ebto probleme dol\u00ebn gjat\u00eb zhvillimit t\u00eb inteligjenc\u00ebs artificiale t\u00eb bazuar n\u00eb m\u00ebsimin e makin\u00ebs p\u00ebr Unleash<\/a><\/noindex>, ku AI duhet t\u00eb sillet si lojtar\u00ebt e zakonsh\u00ebm \u2014 pra, t\u00eb jet\u00eb po aq fleksib\u00ebl dhe shpik\u00ebs. <\/p>\n

Si n\u00eb Starcraft II, Unleashed \u00ebsht\u00eb shum\u00eb m\u00eb e nd\u00ebrlikuar se shahu dhe go p\u00ebr Atari. Procesi i loj\u00ebs \u00ebsht\u00eb intuitiv dhe leht\u00ebsisht i zot\u00ebruari, por p\u00ebr t\u00eb arritur v\u00ebrtet sukses, nevojiten disa aft\u00ebsi menaxhimi t\u00eb metave. Lojtari duhet t\u00eb nd\u00ebrtoj\u00eb labirinte, t\u00eb d\u00ebrgoj\u00eb monsters mbi armiqt\u00eb dhe t\u00eb zhvilloj\u00eb strategjin\u00eb e tij p\u00ebr ekonomin\u00eb, sulmin dhe mbrojtjen e strukturave gjat\u00eb gjith\u00eb loj\u00ebs. P\u00ebr k\u00ebt\u00eb, ai duhet t\u00eb p\u00ebrg\u00ebnjeshtroj\u00eb dhe t\u00eb parashikoj\u00eb l\u00ebvizjet e t\u00eb tjer\u00ebve, si dhe t\u00eb menaxhoj\u00eb metafik\u00ebn psikologjike \u2014 pik\u00ebrisht ajo e b\u00ebn pokerin di\u00e7ka m\u00eb shum\u00eb se nj\u00eb loj\u00eb statistike.<\/p>\n

\"Si
\nScreenshot nga Unleashed<\/i><\/p>\n

N\u00eb k\u00ebrkim t\u00eb arkitektur\u00ebs m\u00eb t\u00eb p\u00ebrshtatshme p\u00ebr k\u00ebto q\u00ebllime, n\u00eb fillim loj\u00ebs i u hap\u00ebn teknologjit\u00eb si evolucioni neural<\/a><\/noindex> dhe m\u00ebsimi i thell\u00eb, dhe u verifikua se si do t\u00eb paraqiteshin n\u00eb form\u00eb t\u00eb pap\u00ebrpunuar si nj\u00eb AI kund\u00ebrshtar. <\/p>\n

Ishte e tmerrshme. <\/p>\n

Shpejt u b\u00eb e qart\u00eb se n\u00eb Unleashed duhet t\u00eb zgjidhen shum\u00eb probleme globale, p\u00ebr t\u00eb cilat \u00ebsht\u00eb e v\u00ebshtir\u00eb t\u00eb p\u00ebrshtatet m\u00ebsimi i makin\u00ebs. <\/p>\n

Nj\u00eb prej tyre \u00ebsht\u00eb nd\u00ebrtimi i nj\u00eb labirinti efektiv. Si n\u00eb shum\u00eb loj\u00ebra q\u00eb kan\u00eb si q\u00ebllim mbrojtjen e kull\u00ebs, k\u00ebtu lojtar\u00ebt duhet t\u00eb nd\u00ebrtoshin nj\u00eb labirint p\u00ebrreth saj, p\u00ebrmes t\u00eb cilit do t\u00eb kalojn\u00eb monsters. Ata, nga ana e tyre, duhet t\u00eb eliminohen me arm\u00ebt e vendosura n\u00eb labirint. Idealisht, labirinti duhet t\u00eb jet\u00eb sa m\u00eb i gjat\u00eb p\u00ebr t\u00eb mundur t\u00eb d\u00ebmtoj\u00eb t\u00eb mjaftuesh\u00ebm monsters dhe p\u00ebr t\u00eb mos lejuar ata t\u00eb arrijn\u00eb tek kulla. P\u00ebr disa lloje arm\u00ebsh, monsters jan\u00eb m\u00eb t\u00eb ndjesh\u00ebm se p\u00ebr t\u00eb tjerat, prandaj p\u00ebr m\u00eb shum\u00eb efikasitet, ato duhet t\u00eb vendosen n\u00eb labirint para t\u00eb tjerave. Ve\u00e7oria e Unleash \u00ebsht\u00eb se nuk ekziston nj\u00eb labirint ideal: n\u00eb loj\u00eb ka kaq shum\u00eb lloje monsters q\u00eb ndonj\u00ebher\u00eb dikush nga ata do t\u00eb kaloj\u00eb pa pengesa n\u00ebp\u00ebr ndonj\u00eb pjes\u00eb t\u00eb labirintit. \u00c7do labirint duhet t\u00eb p\u00ebrshtatet p\u00ebr monsters e reja q\u00eb l\u00ebshohen nga lojtar\u00eb t\u00eb tjer\u00eb. Prandaj, ishte e nevojshme t\u00eb m\u00ebsohej inteligjenca artificiale jo vet\u00ebm t\u00eb nd\u00ebrtonte labirinte, por t\u00eb m\u00ebsohej t\u00eb krijonte labirinte efektive p\u00ebr skenar\u00eb t\u00eb ndrysh\u00ebm, me t\u00eb cilat mund t\u00eb p\u00ebrballej si n\u00eb versionin e hersh\u00ebm ashtu edhe n\u00eb at\u00eb t\u00eb vonsh\u00ebm t\u00eb loj\u00ebs. <\/p>\n

Po t\u00eb nj\u00ebjt\u00ebn m\u00ebnyr\u00eb, AI duhet t\u00eb m\u00ebsoj\u00eb t\u00eb parashikoj\u00eb se cil\u00ebt monstra do t\u00eb shfaqen n\u00eb labirint. Kjo \u00ebsht\u00eb nj\u00eb lloj problemi i kund\u00ebrt n\u00eb krahasim me nd\u00ebrtimin e labirintit. Ashtu si n\u00eb shum\u00eb loj\u00ebra t\u00eb tjera, n\u00eb Unleash nuk mjafton thjesht t\u00eb rrit\u00ebsh ushtrin\u00eb dhe ta d\u00ebrgosh at\u00eb n\u00eb kampin armik: \u00ebsht\u00eb gjithashtu e nevojshme t\u00eb spiunosh mbrojtjen armike dhe t\u00eb strukturosh ushtrin\u00eb n\u00eb m\u00ebnyr\u00eb q\u00eb t\u00eb godas\u00ebsh sa m\u00eb efektivisht pikat e dob\u00ebta t\u00eb kund\u00ebrshtarit. Ushtria e monstruve duhet t\u00eb bashk\u00ebpunoj\u00eb n\u00eb m\u00ebnyr\u00eb q\u00eb t\u00eb kaloj\u00eb me sukses p\u00ebrmes labirintit. Ndonj\u00ebher\u00eb \u00ebsht\u00eb e nevojshme t\u00eb l\u00ebshesh monstrat n\u00eb rend t\u00eb caktuar n\u00eb var\u00ebsi t\u00eb funksioneve dhe rolit t\u00eb tyre. Kjo gjithashtu rrit numrin e kombinimeve t\u00eb ndryshme.<\/p>\n

M\u00eb n\u00eb fund, duke qen\u00eb se lojtari duhet t\u00eb krijoj\u00eb labirintet dhe t\u00eb mbledh\u00eb ushtrin\u00eb e monstruve, AI gjithashtu duhet t\u00eb m\u00ebsoj\u00eb t\u00eb gjej\u00eb nj\u00eb balanc\u00eb midis sulmit dhe mbrojtjes. Duhet gjithashtu t\u00eb merren parasysh q\u00eb sa m\u00eb shum\u00eb q\u00eb lojtari rrit ushtrin\u00eb e monstruve dhe sa m\u00eb shum\u00eb nd\u00ebrtuar labirint, aq m\u00eb shum\u00eb resurse i nevojiten atij. Prandaj, nj\u00eb strategji e duhur sulmi \u00ebsht\u00eb jasht\u00ebzakonisht e r\u00ebnd\u00ebsishme si p\u00ebr ekonomin\u00eb gjat\u00eb loj\u00ebs ashtu edhe p\u00ebr fitoren. Dhe p\u00ebr t\u00eb siguruar konkurrueshm\u00ebri, AI duhet t\u00eb jet\u00eb i aft\u00eb t\u00eb menaxhoj\u00eb burimet n\u00eb m\u00ebnyr\u00eb q\u00eb t\u00eb krijoj\u00eb nj\u00eb ushtri t\u00eb fuqishme monstrush, pa e rrezikuar forc\u00ebn e labirintit. Investimi maksimal n\u00eb monstra mund t\u00eb jet\u00eb ekonomikisht i dobish\u00ebm, por kjo rrit rrezikun e kapjes s\u00eb labirintit nga monstrut e armikut. N\u00ebse vendosni t\u00eb forconi mbrojtjen e labirintit, kjo mund t\u00eb d\u00ebmtoj\u00eb ekonomin\u00eb tuaj. Asnj\u00eb nga k\u00ebto skenar\u00eb nuk do t\u00eb \u00e7oj\u00eb n\u00eb fitore. K\u00ebshtu, problemi i optimizimit n\u00eb Unleash \u00ebsht\u00eb m\u00eb i gjer\u00eb se n\u00eb rastin e shahut ose Starcraft, dhe p\u00ebrfshin nevoj\u00ebn p\u00ebr t\u00eb sakrifikuar di\u00e7ka dhe p\u00ebr t\u00eb parashikuar p\u00ebrfitimin tuaj disa hapa p\u00ebrpara. <\/p>\n

Me m\u00ebsimin e inteligjenc\u00ebs artificiale, shfaqen shum\u00eb probleme q\u00eb nuk ishin marr\u00eb parasysh m\u00eb par\u00eb. N\u00eb fillim, AI shpesh arrinte nj\u00eb nivel t\u00eb caktuar zhvillimi, ku fillonte t\u00eb kuptonte disa aspekte t\u00eb loj\u00ebs - p\u00ebr shembull, cilat arm\u00eb n\u00eb labirint jan\u00eb efektive kund\u00ebr tipeve t\u00eb caktuara t\u00eb monstruve ose cilat monstru jan\u00eb m\u00eb t\u00eb p\u00ebrshtatsh\u00ebm p\u00ebr t\u00eb kaluar disa pjes\u00eb t\u00eb labirintit. Por m\u00ebsimi ishte i ngadalt\u00eb dhe \u00e7onte n\u00eb krijimin e strategjive monotone. <\/p>\n

<\/p>\n

Nevoja p\u00ebr qasje paralele<\/h3>\n

\nNd\u00ebrsa m\u00ebsimi i AI-ve t\u00eb bazuara n\u00eb m\u00ebsimin e makineris\u00eb po p\u00ebrparonte ngadal\u00eb dhe nuk rezultonte shum\u00eb me sukses, p\u00ebr etapa t\u00eb tjera t\u00eb testimit dhe zhvillimit ishte e nevojshme nj\u00eb AI m\u00eb i mir\u00eb dhe AI konkurruese m\u00eb t\u00eb besueshme. P\u00ebr realizimin e tyre u p\u00ebrdor arkitektura Utility, q\u00eb e mund\u00ebson krijimin e AI-ve t\u00eb ve\u00e7anta p\u00ebr testimin dhe verifikimin e cil\u00ebsis\u00eb s\u00eb loj\u00ebs, testimeve brenda loj\u00ebs dhe balancimit t\u00eb arm\u00ebve dhe monstruve, si dhe krijimin e labirinteve dhe monstruve specifike. Megjithat\u00eb, gjat\u00eb zhvillimit t\u00eb Unleash, krijuesit vet\u00eb pat\u00ebn p\u00ebrmir\u00ebsuar aft\u00ebsit\u00eb e tyre n\u00eb kalimin e saj, dhe m\u00eb pas vendos\u00ebn t\u00eb p\u00ebrdorin njohurit\u00eb e fituara p\u00ebr t\u00eb krijuar nj\u00eb Utility AI m\u00eb t\u00eb avansuar. K\u00ebshtu, u kuptua se shum\u00eb probleme q\u00eb shfaqen n\u00eb sistemet e inteligjenc\u00ebs artificiale t\u00eb bazuara n\u00eb m\u00ebsimin e makineris\u00eb mund t\u00eb zgjidhen leht\u00ebsisht me ndihm\u00ebn e sistemeve Utility q\u00eb p\u00ebrdorin njohurit\u00eb e vendosura n\u00eb to, dhe anasjelltas. <\/p>\n

P\u00ebr shembull, labirintet m\u00eb efektive \u00ebsht\u00eb m\u00eb mir\u00eb t'i nd\u00ebrtosh me an\u00eb t\u00eb AI Utility, t\u00eb bazuara n\u00eb bazat e njohurive t\u00eb formuara nga rezultatet e testeve t\u00eb brendshme. \u00cbsht\u00eb e leht\u00eb t\u00eb p\u00ebrshkruash dhe programosh nj\u00eb algorit\u00ebm p\u00ebr nd\u00ebrtimin e labirintit dhe vendosjen e arm\u00ebve n\u00eb t\u00eb n\u00eb m\u00ebnyr\u00eb q\u00eb t\u00eb ishte m\u00eb e leht\u00eb p\u00ebr nj\u00eb lojtar t\u00eb mbroj\u00eb kull\u00ebn nga monstru t\u00eb caktuar. Por krijimi i nj\u00eb ushtrie monstru n\u00eb baz\u00eb t\u00eb njohurive p\u00ebr baz\u00ebn e armikut \u00ebsht\u00eb nj\u00eb detyr\u00eb e komplikuar p\u00ebr k\u00ebto AI, pasi numri i kushteve dhe kombinimeve q\u00eb duhet t\u00eb merren parasysh ishte i friksh\u00ebm. Me k\u00ebt\u00eb lloj arkitekture, k\u00ebrkimi i grupeve t\u00eb p\u00ebrshtatshme t\u00eb monstruve do t\u00eb zgjaste pafund\u00ebsisht. N\u00eb k\u00ebto rrethana, e m\u00ebsuara e thell\u00eb do t\u00eb ishte zgjidhja ideale p\u00ebr k\u00ebt\u00eb detyr\u00eb. <\/p>\n

<\/p>\n

Krijimi i AI-ve hibride<\/h3>\n

\nPraefat, u deshtua q\u00eb t\u00eb bashkohen dy qasje dhe k\u00ebshtu t\u00eb krijohet nj\u00eb sistem hibrid t\u00eb inteligjenc\u00ebs artificiale t\u00eb bazuar n\u00eb m\u00ebsimin e makinerive dhe Utility. Ideja ishte q\u00eb atje ku ishte e nevojshme t\u00eb p\u00ebrpunoheshin sasi t\u00eb m\u00ebdha kombinimesh dhe gjendjesh t\u00eb loj\u00ebs, ose ku k\u00ebrkohej q\u00eb ajo t\u00eb m\u00ebsohej di\u00e7ka, t\u00eb p\u00ebrdorej m\u00ebsimi i makinerive. P\u00ebr detyra t\u00eb tjera, ku ishte m\u00eb mir\u00eb t\u00eb mb\u00ebshteteshim n\u00eb p\u00ebrvoj\u00ebn personale t\u00eb zhvilluesve, u angazhuan sistemet Utility. Avantazhi i k\u00ebsaj qasje \u00ebsht\u00eb gjithashtu se, n\u00ebse nevojitet, mund t\u00eb kontrollohet m\u00eb mir\u00eb sjellja e AI-s\u00eb p\u00ebr t\u00eb siguruar q\u00eb ajo t\u00eb ndjek\u00eb sakt\u00ebsisht q\u00ebllimin e caktuar. P\u00ebr shembull, mund t\u00eb p\u00ebrdoret Utility AI p\u00ebr t\u00eb siguruar ekuilibrin midis sulmit dhe mbrojtjes dhe k\u00ebshtu t\u00eb krijohen nivele t\u00eb ndryshme agresiviteti, ose mund t\u00eb krijohen konfiguracione t\u00eb ndryshme labirintesh p\u00ebr AI t\u00eb ndryshme, p\u00ebr t\u00eb formuar p\u00ebr to stile individuale t\u00eb loj\u00ebs. Gjithashtu, mund t\u00eb caktohen rrjeteve nervore disa sisteme vlerash p\u00ebr t\u00eb formuar preferenca t\u00eb ndryshme n\u00eb p\u00ebrzgjedhjen e monstrave ajrore ose tok\u00ebsore dhe n\u00eb k\u00ebt\u00eb m\u00ebnyr\u00eb t\u00eb p\u00ebrshtaten individualitetet p\u00ebr AI t\u00eb ve\u00e7anta. Ka shum\u00eb mund\u00ebsi t\u00eb tjera p\u00ebr realizimin e zgjidhjeve projektore, dhe t\u00eb gjitha ato n\u00ebnvizojn\u00eb pikat e forta t\u00eb \u00e7do lloj arkitekture t\u00eb AI-s\u00eb. <\/p>\n

Qasja hibride gjithashtu ishte p\u00ebrgjigje ndaj nj\u00eb pyetje tjet\u00ebr q\u00eb u paraqit p\u00ebr ekipin e zhvilluesve gjat\u00eb procesit t\u00eb zhvillimit t\u00eb inteligjenc\u00ebs artificiale p\u00ebr Unleash: a duhet t\u00eb aplikojm\u00eb nj\u00eb rrjet t\u00eb thell\u00eb global t\u00eb vog\u00ebl t\u00eb bazuar n\u00eb m\u00ebsimin e makinerive p\u00ebr t\u00eb marr\u00eb parasysh t\u00eb dh\u00ebnat hyr\u00ebse dhe dal\u00ebse, apo \u00ebsht\u00eb m\u00eb mir\u00eb t\u00eb projektosh AI me nj\u00eb struktur\u00eb hierarkike? <\/p>\n

\"Si
\nDy arkitekturat e p\u00ebrdorura n\u00eb Unleash: majtas \u2014 nj\u00eb rrjet i madh i thell\u00eb me nj\u00eb arkitektur\u00eb t\u00eb vetme, djathtas \u2014 nj\u00eb sistem hierarkik ku \u00e7do rrjet ka detyr\u00ebn e vet.<\/i><\/p>\n

Megjithat\u00eb, ishte e d\u00ebshirueshme t\u00eb krijohej nj\u00eb qasje e p\u00ebrgjithshme p\u00ebr sistemin e inteligjenc\u00ebs artificiale, n\u00eb arkitektur\u00ebn e t\u00eb cilit zhvilluesit nuk do t\u00eb inkuadrojn\u00eb p\u00ebrvoj\u00ebn e tyre personale. Megjithat\u00eb, sa m\u00eb shum\u00eb hyrje t\u00eb b\u00ebheshin n\u00eb loj\u00eb, aq m\u00eb tep\u00ebr rritej rrjeti nervor. N\u00eb k\u00ebt\u00eb rast, nuk ishte e mundur t\u00eb ndaheshin m\u00ebsimi i AI-s\u00eb dhe t\u00eb m\u00ebsonin di\u00e7ka t\u00eb vetme: apo mbrojtjen, apo sulmin. Dhe shfaqeshin shqet\u00ebsime se nj\u00eb qasje m\u00eb e p\u00ebrgjithshme do t\u00eb \u00e7onte n\u00eb nj\u00eb rritje t\u00eb konsiderueshme t\u00eb numrit t\u00eb llogaritjeve. <\/p>\n

Pra, lindi ideja p\u00ebr t\u00eb krijuar nj\u00eb arkitektur\u00eb hierarkike, n\u00eb t\u00eb cil\u00ebn \u00e7do detyr\u00eb specifike do t\u00eb realizohej nga nj\u00eb rrjet i specializuar nervor. Sipas k\u00ebsaj ideje, fillimisht inteligjenca artificiale duhet t\u00eb marr\u00eb nj\u00eb vendim p\u00ebr shp\u00ebrndarjen e burimeve n\u00eb sulm (rritjen e ushtris\u00eb s\u00eb monstrave) dhe mbrojtje (nd\u00ebrtimin e labirintit). Pik\u00ebrisht pasi b\u00ebn k\u00ebt\u00eb, kalon n\u00eb katin tjet\u00ebr sipas zgjedhjes s\u00eb tij dhe merr akses n\u00eb pjes\u00ebn e nevojshme t\u00eb gjendjes s\u00eb loj\u00ebs, pas s\u00eb cil\u00ebs merr vendime t\u00eb detajuara p\u00ebr se cilat monst\u00ebr t\u00eb zgjedh\u00eb dhe \u00e7far\u00eb arme t\u00eb vendos\u00eb n\u00eb labirint.<\/p>\n

<\/p>\n

P\u00ebrfundimi dhe hapat e ardhsh\u00ebm<\/h3>\n

\nMe nj\u00eb qasje hibride, Utility AI me rrjetet e tij t\u00eb bazuara n\u00eb m\u00ebsimin e makinerive ngjan me arkitektur\u00ebn hierarkike. Dhe kjo, nga ana e saj, \u00ebsht\u00eb e ngjashme me trurin biologjik, ku qendrat e ndryshme nervore p\u00ebrgjigjen p\u00ebr \u00e7do detyr\u00eb.<\/p>\n

Tani, AI i armikut n\u00eb Unleash \u00ebsht\u00eb shum\u00eb e v\u00ebshtir\u00eb p\u00ebr t'u mposhtur: ata jan\u00eb n\u00eb gjendje t\u00eb p\u00ebrshtaten me \u00e7do situat\u00eb n\u00eb loj\u00eb, \u2014 por megjithat\u00eb zhvilluesit mund t\u00eb ndryshojn\u00eb parametrat e tyre sipas d\u00ebshirs. Sipas autorit t\u00eb artikullit, me kalimin e koh\u00ebs, qasja hibride duhet t\u00eb p\u00ebrfshij\u00eb m\u00eb shum\u00eb dhe t\u00eb shfaqet n\u00eb shum\u00eb loj\u00ebra t\u00eb tjera. Ndoshta nj\u00eb dit\u00eb do t\u00eb jet\u00eb e mundur t\u00eb integrohet nj\u00eb inteligjenc\u00eb artificiale e bazuar n\u00eb m\u00ebsimin e makinerive n\u00eb m\u00ebnyr\u00eb t\u00eb past\u00ebr n\u00eb procesin e loj\u00ebs. Por, \u00ebsht\u00eb e qart\u00eb se p\u00ebr k\u00ebt\u00eb do t\u00eb nevojitet ende koh\u00eb. P\u00ebr momentin, q\u00ebllimi \u00ebsht\u00eb t\u00eb gjendet nj\u00eb arkitektur\u00eb e till\u00eb q\u00eb p\u00ebrshtatet vet\u00eb me detyrat q\u00eb ka p\u00ebrpara dhe gjen m\u00ebnyra optimale p\u00ebr t'i zgjidhur ato.
\n
Burimi: habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u0412 \u043f\u0440\u043e\u0434\u043e\u043b\u0436\u0435\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0433\u0434\u0430 \u043f\u043e\u0434\u043d\u044f\u0442\u043e\u0439 \u0432 \u043d\u0430\u0448\u0435\u043c \u0431\u043b\u043e\u0433\u0435 \u0442\u0435\u043c\u044b \u0438\u0433\u0440\u043e\u0432\u043e\u0433\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430 \u043f\u043e\u0433\u043e\u0432\u043e\u0440\u0438\u043c \u043e \u0442\u043e\u043c, \u043d\u0430\u0441\u043a\u043e\u043b\u044c\u043a\u043e \u043f\u0440\u0438\u043c\u0435\u043d\u0438\u043c\u043e \u043a \u043d\u0435\u043c\u0443 \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u0438 \u0432 \u043a\u0430\u043a\u043e\u043c \u0432\u0438\u0434\u0435. \u0421\u0432\u043e\u0438\u043c \u043e\u043f\u044b\u0442\u043e\u043c \u0438 \u0432\u044b\u0431\u0440\u0430\u043d\u043d\u044b\u043c\u0438 \u043d\u0430 \u0435\u0433\u043e \u043e\u0441\u043d\u043e\u0432\u0435 \u0440\u0435\u0448\u0435\u043d\u0438\u044f\u043c\u0438 \u043f\u043e\u0434\u0435\u043b\u0438\u043b\u0441\u044f \u044d\u043a\u0441\u043f\u0435\u0440\u0442 \u043f\u043e \u0432\u043e\u043f\u0440\u043e\u0441\u0430\u043c \u0418\u0418 \u0432 Apex Game Tools \u042f\u043a\u043e\u0431 \u0420\u0430\u0441\u043c\u0443\u0441\u0441\u0435\u043d. \u0412 \u043f\u043e\u0441\u043b\u0435\u0434\u043d\u0438\u0435 \u0433\u043e\u0434\u044b \u0432\u0435\u0434\u0451\u0442\u0441\u044f \u043c\u043d\u043e\u0433\u043e \u0440\u0430\u0437\u0433\u043e\u0432\u043e\u0440\u043e\u0432 \u043e \u0442\u043e\u043c, \u0447\u0442\u043e \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u043a\u0430\u0440\u0434\u0438\u043d\u0430\u043b\u044c\u043d\u043e […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-52119","post","type-post","status-publish","format-standard","hentry","category-novosti-interneta"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Si funksionon AI hibrid i loj\u00ebrave dhe cfar\u00eb p\u00ebrpar\u00ebsish ka | ProHoster","description":"N\u00eb vazhdim t\u00eb tem\u00ebs s\u00eb dikurshme t\u00eb zgjidhur n\u00eb blogun ton\u00eb mbi inteligjenc\u00ebn artificiale n\u00eb loj\u00ebra, do t\u00eb flasim se sa e aplikueshme \u00ebsht\u00eb.","canonical_url":"https:\/\/prohoster.info\/sq\/blog\/novosti-interneta\/kak-ustroen-gibridnyj-igrovoj-ii-i-v-chyom-ego-preimushhestva","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sq_AL","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47\u041a\u0430\u043a \u0443\u0441\u0442\u0440\u043e\u0435\u043d \u0433\u0438\u0431\u0440\u0438\u0434\u043d\u044b\u0439 \u0438\u0433\u0440\u043e\u0432\u043e\u0439 \u0418\u0418 \u0438 \u0432 \u0447\u0451\u043c \u0435\u0433\u043e \u043f\u0440\u0435\u0438\u043c\u0443\u0449\u0435\u0441\u0442\u0432\u0430 | ProHoster","og:description":"\u0412 \u043f\u0440\u043e\u0434\u043e\u043b\u0436\u0435\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0433\u0434\u0430 \u043f\u043e\u0434\u043d\u044f\u0442\u043e\u0439 \u0432 \u043d\u0430\u0448\u0435\u043c \u0431\u043b\u043e\u0433\u0435 \u0442\u0435\u043c\u044b \u0438\u0433\u0440\u043e\u0432\u043e\u0433\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430 \u043f\u043e\u0433\u043e\u0432\u043e\u0440\u0438\u043c \u043e \u0442\u043e\u043c, \u043d\u0430\u0441\u043a\u043e\u043b\u044c\u043a\u043e \u043f\u0440\u0438\u043c\u0435\u043d\u0438\u043c\u043e.","og:url":"https:\/\/prohoster.info\/sq\/blog\/novosti-interneta\/kak-ustroen-gibridnyj-igrovoj-ii-i-v-chyom-ego-preimushhestva","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-10-31T21:00:00+00:00","article:modified_time":"2020-02-18T10:59:47+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"52119","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-24 02:32:21","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 20:49:49","updated":"2026-01-24 02:32:21","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/52119","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/comments?post=52119"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/52119\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/media?parent=52119"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/categories?post=52119"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/tags?post=52119"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}