Zviri Mukati[Viga][Ratidza]
Pfungwa yekuti marobhoti akangwara kupfuura vanhu yakatora fungidziro yedu pamwe chete kwenguva yakareba sekunge kwave neScience Fiction.
Nekudaro, nepo Artificial Intelligence (AI) isati yasvika padanho iro, isu takaita budiriro huru mukugadzira hungwaru hwemuchina, sezvakaratidzwa neGoogle, Tesla, uye Uber kuyedza nemotokari dzinozvityaira.
Kusarudzika uye kushandiswa kweKudzidza Kwakadzika, iyo Muchina Kudzidza nzira iyo inogonesa kufambira mberi kwehunyanzvi, ine chikamu chebasa rekubudirira kweAI kubva kumayunivhesiti uye marabhoritari ekutsvagisa kuenda kune zvigadzirwa.
Iyo inotevera komputa shanduko ichavakwa pahuchenjeri hwekugadzira, kudzidza kwakadzama, uye machine learning.
Aya matekinoroji akavakirwa pakugona kuona mapatani uyezve kufanotaura zviitiko zvenguva yemberi zvichienderana nedata rakaunganidzwa munguva yakapfuura. Izvi zvinotsanangura chikonzero nei Amazon ichiita mazano kana uchitenga pamhepo kana kuti Netflix inokuziva sei iwe unofarira anotyisa 1980s mafirimu.
Kunyange zvazvo makombiyuta anoshandisa pfungwa dzeAI dzimwe nguva anonzi "smart," ruzhinji rwezvirongwa izvi hazvidzidziri pachavo; kudyidzana kwevanhu kunodiwa.
Masayendisiti edata anogadzirira mapeji nekutora machinjiro anozoiswa mukati predictive analytics. Kudzidza kwakadzama, kune rumwe rutivi, kunogona kuita basa iri otomatiki.
Iyi posvo inoshanda segwara remunda kune chero vanoda data vanofarira kudzidza zvakawanda nezve kudzidza kwakadzama, hupamhi hwayo, uye ramangwana rinogona kuitika.
Chii chinonzi Deep Learning?
Kudzidza kwakadzama kunogona kufungidzirwa sechikamu chekudzidza kwemichina.
Imunda wakavakirwa pakudzidza wega uye kuvandudza kuburikidza nekuongorora makomputa algorithms.
Kudzidza kwakadzama, kusiyana nekudzidza kwemuchina, kunoshanda neyakagadzirwa neural networks, izvo zvinofanirwa kutevedzera mafungire nekudzidza vanhu. Kusvika nguva pfupi yadarika, neural network yaive yakaganhurirwa mukuomarara nekuda kwekunetsa kwemagetsi emakomputa.
Nekudaro, kufambira mberi muBig Data analytics kwagonesa akakura, ane simba neural network, zvichiita kuti makomputa atarise, anzwisise, uye apindure kumamiriro akaoma nekukurumidza kupfuura vanhu.
Tinokurudzira kuverenga - Tesla Neural Network Architecture Inotsanangurwa
Kukamurwa kwemifananidzo, kushandura mutauro, uye kuzivikanwa kwekutaura zvese zvakabatsirikana nekudzidza kwakadzama. Inogona kugadzirisa chero nyaya yekuzivikanwa kwepateni pasina kudiwa kwekudyidzana kwevanhu.
Izvo zvine matatu- kana kupfuura-layered neural network. Aya neural network anotsvaga kutevedzera mashandiro ehuropi hwemunhu, kunyangwe nekubudirira kushoma, nekuigonesa "kudzidza" kubva kune yakakura mavhoriyamu edata.
Nepo chidimbu chimwe chete cheneural network chichigona kuburitsa fungidziro, mamwe akavigwa akaturikidzana anogona kubatsira mukugadzirisa uye kugadzirisa kwechokwadi.
Chii chinonzi Neural Network?
Artificial neural network yakavakirwa pane neural nets inoonekwa muuropi hwemunhu. Kazhinji, neural network inoumbwa nezvikamu zvitatu.
Iwo matatu mazinga ndeokupinda, kubuda, uye akavanzwa. A neural network in action inoonekwa mumufananidzo uri pazasi.
Sezvo iyo neural network inoratidzwa pamusoro ichingove neimwe yakavanzika layer, inonzi "shallow neural network."
Mamwe matinji akavanzika anowedzerwa kune akadaro masisitimu kuti aumbe mamwe maumbirwo akaomarara.
Chii chinonzi Deep Network?
Mune yakadzika network, akawanda akavanzika akaturikidzana anowedzerwa.
Kudzidzira magadzirirwo akadaro kunowedzera kuomarara sezvo huwandu hwematanho akavanzika mumambure anokwira, kwete chete maererano nenguva inodiwa kudzidzisa zvakanaka network asiwo maererano nezviwanikwa zvinodiwa.
Network yakadzika ine yekupinza, mana akavanzika akaturikidzana, uye inobuda inoratidzwa pazasi.
Kudzidza Kwakadzika kunoshanda sei?
Neural network inovakwa nezvikamu zvenodhi, zvakafanana nemagadzirirwo anoita uropi hwemunhu. Individual layer node dzakabatana nemanodhi muzvikamu zvakavakidzana.
Huwandu hwema layer mune network hunoratidza kudzika kwayo. Imwe neuron mu ubongo hwevanhu inogamuchira zviuru zvemashoko.
Zviratidzo zvinofamba pakati pemanodhi mune artificial neural network, iyo inogovera huremu kwavari.
Node ine huremu hwepamusoro ine simba guru pamanodhi pasi payo. Mutsara wekupedzisira unosanganisa zvakaiswa huremu kuti zvipe kubuda.
Masisitimu ekudzidza akadzama anoda Hardware yakasimba nekuda kwehuwandu hukuru hwe data inobatwa uye kuwanda kwehunyanzvi hwemasvomhu computations inobatanidzwa.
Kuverenga kwakadzama kwekudzidzira kuverenga, kunyangwe nehunyanzvi hwakadaro, kunogona kutora mavhiki.
Dzidzo dzakadzama masisitimu anoda huwandu hwakakosha hwe data kuti ipe zvakawanikwa zvakarurama; saka, ruzivo rwunopihwa muchimiro chedatasets hombe.
Kana uchigadzira data, artificial neural network inogona kurongedza ruzivo zvichienderana nemhinduro kune kutevedzana kwebhinari hongu kana mibvunzo yenhema inosanganisira yakaoma kwazvo masvomhu computations.
Algorithm yekuzivikanwa kwechiso, semuenzaniso, inodzidza kuziva uye kuziva mipendero nemitsetse yezviso.
Zvino zvimwe zvakakosha zvezviso, uye pakupedzisira zvinomiririra zviso.
Iyo algorithm inozvidzidzisa nekufamba kwenguva, ichiwedzera mukana wemhinduro dzakakodzera.
Mumamiriro ezvinhu aya, iyo algorithm yekuzivikanwa kwechiso inozoziva zviso zvakanyanya nekufamba kwenguva.
Kudzidza Kwakadzika VS Muchina Kudzidza
Kudzidza kwakadzama kwakasiyana sei nekudzidza kwemuchina kana chiri chikamu chayo?
Kudzidza kwakadzama kunosiyana kubva pachinyakare muchina kudzidza mumhando dze data yainoshandisa uye nzira dzaanoshandisa kudzidza.
Kugadzira fungidziro, maalgorithms ekudzidza muchina anoshandisa data rakarongeka, rakanyorwa, izvo zvinoreva kuti humwe hunhu hunotsanangurwa kubva kudhata rekupinza remuenzaniso uye rakaiswa mumatafura.
Izvi hazvireve hazvo kuti haishandise data isina kurongeka; asi, kana ikadaro, inowanzoenda kuburikidza nekumwe kusati kwagadziriswa kuti iise muchimiro chakarongeka.
Kudzidza kwakadzama kunobvisa chikamu che data pre-processing iyo kudzidza muchina kunowanzo sanganisira.
Aya maalgorithms anogona kupinza uye kududzira data isina kurongeka senge zvinyorwa nemifananidzo, pamwe nekuita otomatiki kudhirowa, kuderedza kuvimba nenyanzvi dzevanhu.
Ngatimbofungidzirai takanga tine muunganidzwa wemifananidzo yezvipfuyo zvakasiyana zvataida kuronga muzvikamu zvakaita se “katsi,” “imbwa,” “hamster,” zvichingodaro.
Kudzidza kwakadzama algorithms kunogona kuona kuti ndeapi maitiro (senzeve) anonyanya kukosha mukuparadzanisa mhuka kubva kune imwe. Ichi chimiro che hierarchy chinotsanangurwa nemaoko nenyanzvi yemunhu mukudzidza kwemichina.
Iyo yakadzika yekudzidza sisitimu inozochinja uye inokwana pachayo kuti inyatso kuburikidza gradient descent uye kudzosera kumashure, zvichiibvumira kuti ibudise fungidziro dzakanyatsojeka nezvemufananidzo mutsva wemhuka.
Kudzidza kwakadzama zvikumbiro
1 Chatbots
Chatbots inogona kugadzirisa nyaya dzevatengi mune imwe nyaya yemasekonzi. Chatbot ndeye chakagadzirwa njere (AI) chishandiso chinokutendera kuti utaure online kuburikidza nemavara kana mameseji-kune-kutaura.
Inogona kutaurirana uye kuita zviito nenzira imwechete iyo vanhu vanoita. Chatbots anoshandiswa zvakanyanya musevhisi yevatengi, social media kushambadzira, uye vatengi kutumira meseji.
Inopindura zvaunopinza nemhinduro otomatiki. Inogadzira nzira dzakawanda dzemhinduro uchishandisa muchina kudzidza uye hunyanzvi hwekudzidza hwakadzama.
2. Mota dzinozvifambisa
Kudzidza Kwakadzika ndicho chinhu chikuru chinoita kuti mota dzinozvityaira dzive dzechokwadi.
Mamiriyoni e data seti akaiswa muhurongwa kugadzira modhi, dzidzisa michina kudzidza, uye wobva waongorora zvakawanikwa munzvimbo yakachengeteka.
Iyo Uber chakagadzirwa Intelligence Labs muPittsburgh haisi kungoyedza kuita kuti mota dzisina mutyairi dziwande asi kubatanidza akawanda akangwara maficha, senge mikana yekuunza chikafu, nekushandiswa kwemotokari dzisingatyairi.
Kunyanya kunetseka kwekuzvifambisa-kutyaira kwemotokari kutarisana nezviitiko zvisingatarisirwi.
Kutenderera kutenderera kwekuyedza uye kuita, kwakajairwa kwakadzama kudzidza algorithms, kunovimbisa kutyaira kwakachengeteka sezvo kunoratidzwa kune mamirioni ezviitiko zvakanyanya uye zvakanyanya.
3. Virtual Mubatsiri
MaVirtual Assistants mapurogiramu-akavakirwa makore anoziva mirairo yemutauro wechisikigo uye anoita zvinhu panzvimbo yako.
Vabatsiri veVirtual vakaita seAmazon Alexa, Cortana, Siri, uye Google Mubatsiri mienzaniso yakajairika.
Kuti vashandise zvizere zvavanogona, vanoda michina yakabatana neinternet. Kana murairo unopiwa kumubatsiri, zvinowanzo kuendesa chiitiko chiri nani kubva pane zvakambosangana nazvo uchishandisa Deep Kudzidza algorithms.
4. Varaidzo
Makambani akaita seNetflix, Amazon, YouTube, uye Spotify anopa akakodzera bhaisikopo, rwiyo, uye vhidhiyo mazano kune vatengi vavo kuti vavandudze ruzivo rwavo.
Kudzidza kwakadzama ndiko kunokonzeresa zvese izvi.
Mafemu ekutepfenyura epamhepo anopa zvigadzirwa uye masevhisi kurudziro zvichienderana nenhoroondo yekubhurawuza yemunhu, zvaanofarira, uye chiitiko.
Kudzika kudzidza algorithms anoshandiswawo kugadzira zvinyorwa zvidiki otomatiki uye kuwedzera ruzha kumabhaisikopo akanyarara.
5. Robotic
Kudzidza Kwakadzika kunoshandiswa zvakanyanya mukugadzira marobhoti anogona kuita mabasa akafanana nevanhu.
Marobhoti Akadzika Kudzidza-ane simba anoshandisa zvigadziriso zvenguva chaiyo kuona zvipingamupinyi munzira yavo uye nekukurumidza kuronga nzira yavo.
Inogona kushandiswa kutakura zvinhu muzvipatara, mafekitori, matura, maneja manejimendi, kugadzira zvigadzirwa, zvichingodaro.
Marobhoti eBoston Dynamics anopindura vanhu kana vachisundwa. Vanogona kudurura muchina wekugezesa ndiro, vanogona kusimuka pavanenge vadonha, uye vanogona kuita mamwe mabasa akasiyana siyana.
6. Nezveutano
Vanachiremba havagone kuve nevarwere vavo nguva dzese, asi chinhu chimwe chatinogara tiinacho mafoni edu.
Kudzidza kwakadzama kunobvumirawo matekinoroji ekurapa kuongorora data kubva pamifananidzo yatinotora uye data rekufamba kuti tiburitse zvingangove zvinonetsa pautano.
AI's komputa yekuona chirongwa, semuenzaniso, inoshandisa iyi data kutevedzera mafambiro emurwere kufanotaura kudonha pamwe nekuchinja kwemamiriro epfungwa.
Kudzidza kwakadzama kwakashandiswa zvakare kuona gomarara reganda uchishandisa mapikicha nezvimwe zvakawanda.
7. Natural Language Processing
Kugadzira tekinoroji yekugadzira mitauro yakagonesa marobhoti kuverenga kutaurirana uye kuwana zvinoreva kubva kwavari.
Zvakadaro, matauriro acho anogona kurerutswa, achitadza kuverengera nzira iyo mazwi anobatanidzwa nadzo kukanganisa zvinoreva kana chinangwa chechirevo.
Kudzidza kwakadzama kunobatsira magadzirirwo emutauro wechisikigo kuti aone mapatani akaomarara mumitsara uye ape kududzira kwakaringana.
8. Computer Vision
Kudzidza kwakadzama kunoedza kudzokorora mashandisiro anoita pfungwa dzemunhu ruzivo uye nekuziva mapatani, zvichiita kuti ive nzira yakanaka yekudzidzisa chiono-chakavakirwa AI zvikumbiro.
Iwo masisitimu anogona kutora munhevedzano yeakamaka mafoto seti uye kudzidza kuziva zvinhu zvakaita sendege, zviso, uye zvombo uchishandisa. mienzaniso yekudzidza yakadzama.
Kudzidza Kwakadzika Mukuita
Kunze kweyaunofarira mimhanzi yekufambisa sevhisi inokurudzira nziyo dzaungade, kudzidza kwakadzama kuri kuchinja sei hupenyu hwevanhu?
Kudzidza kwakadzama, zvinozoitika, kuri kuita nzira yayo mukuwanda kwekushandisa. Chero ani anoshandisa Facebook anozoona kuti iwe paunotumira mifananidzo mitsva, saiti yemagariro inowanzo ziva uye inomaka shamwari dzako.
Kudzidza kwakadzama kunoshandiswa kugadzirisa mutauro wechisikigo uye kuzivikanwa kwekutaura nevabatsiri vedhijitari vakaita saSiri, Cortana, Alexa, uye Google Zvino.
Shanduro yenguva chaiyo inopiwa kuburikidza ne Skype. Mazhinji masevhisi eemail akafambira mberi mukukwanisa kwavo kuona spam meseji vasati vasvika kubhokisi rekutumira.
PayPal yakashandisa kudzidza kwakadzama kudzivirira kubhadhara kwechitsotsi. CamFind, semuenzaniso, inobvumidza iwe kutora foto yechero chinhu uye, uchishandisa nharembozha yekutsvaga tekinoroji, sarudza kuti chii.
Kudzidza kwakadzama kuri kushandiswa kupa mhinduro neGoogle kunyanya. AlphaGo, chirongwa chekombuta chakagadziridzwa neGoogle Deepmind, chaunza Go shasha dzazvino.
WaveNet, yakagadziridzwa neDeepMind, inogona kugadzira kutaura kunonzwika sechisikigo kupfuura mataurirwo aripo parizvino. Kushandura mitauro yemuromo neyezvinyorwa, Google Translate inoshandisa kudzidza kwakadzama uye kuziva mapikicha.
Chero mufananidzo unogona kuzivikanwa uchishandisa Google Planet. Kubatsira mukugadzira maapplication eAI, Google yakagadzira iyo Tensorflow kudzidza kwakadzama software database.
Ramangwana Rekudzidza Kwakadzika
Kudzidza kwakadzama inyaya isingadzivisiki kana tichikurukura nezve tekinoroji. Hazvina basa kutaura, kudzidza kwakadzama kwakashanduka kuita chimwe chezvinhu zvakakosha zvetekinoroji.
Masangano aimbove ega aifarira matekinoroji seAI, kudzidza kwakadzama, kudzidza muchina, zvichingodaro. Vanhu, zvakare, vari kufarira chinhu ichi chetekinoroji, kunyanya kudzidza kwakadzama.
Chimwe chezvikonzero zvakawanda kudzidza kwakadzama kuri kuwana kutariswa kwakanyanya kugona kwayo kubvumidza zvirinani-zvinofambiswa nedata uku uchivandudza chokwadi chekufungidzira.
Maturusi ekuvandudza kudzidza kwakadzama, maraibhurari, uye mitauro inogona kunyatso kuve zvikamu zvenguva dzose zvechero software yekuvandudza maturusi mumakore mashoma.
Aya maturusi emazuva ano seti anovhura nzira yekugadzira nyore, kuseta, uye kudzidziswa kwemamodheru matsva.
Shanduko yemaitiro, auto-tagging, kusikwa kwemimhanzi, uye mamwe mabasa aizova nyore zvikuru kuita nounyanzvi uhu.
Iko kudiwa kwekukurumidza kukodha hakuna kumbove kwakakura.
Vagadziri vekudzidza kwakadzama vanozowedzera kushandisa zvakabatanidzwa, zvakavhurika, makore-based budiriro nharaunda inobvumira kupinda kune dzakasiyana siyana dzekunze-pasherufu uye pluggable algorithm raibhurari mune ramangwana.
Kudzidza kwakadzama kune ramangwana rakajeka!
Kubatsira kwe a neural network ndeyokuti inopfuura pakubata nehuwandu hwemashoko akasiyana-siyana (funga nezvezvinhu zvose ubongo hwedu hunofanira kutarisana nazvo, nguva dzose).
Izvi zvinonyanya kuitika muzera redu reane simba akangwara masensa, anogona kuunganidza huwandu hukuru hwe data. Masisitimu emakombuta echinyakare ari kunetseka kupepeta, kuisa mumapoka, uye kuwana mhedziso kubva kune data rakawanda.
mhedziso
Kudzidza zvakadzika masimba mazhinji eiyo artificial intelligence (AI) mhinduro dzinogona kuvandudza otomatiki uye analytical maitiro.
Vanhu vazhinji vanosangana nekudzidza kwakadzama zuva nezuva kana vachishandisa internet kana nharembozha.
Kudzidza kwakadzama kunoshandiswa kugadzira zvinyorwa zveYouTube mavhidhiyo. Ita kuzivikanwa kwezwi pamafoni uye vatauri vakangwara.
Ipa chitarisiko chechiso chemifananidzo, uye bvumidza mota dzinozvityaira, pakati pezvimwe zvakawanda zvinoshandiswa.
Uye, sezvo data masayendisiti uye vadzidzi vezvidzidzo vanotarisana nekuwedzera kuoma mapurojekiti ekudzidza kwakadzama achishandisa masisitimu ekudzidza akadzama, uhwu hungwaru hwekugadzira huchava chinhu chinowedzera kukosha chehupenyu hwedu hwezuva nezuva.
Leave a Reply