I-AI inceda ukufunda ngezilwanyana e-Afrika

Ukusuka kuyo nayiphi na iketile yombane eqhagamshelwe kwi-Intanethi, unokuva malunga nendlela i-AI ebetha ngayo iimbaleki ze-cyber, inika amathuba amatsha kubuchwephesha obudala, kwaye izoba iikati ngokusekwe kumzobo wakho. Kodwa bathetha kancinci malunga nento yokuba ubukrelekrele bomatshini bukwalawula ukukhathalela okusingqongileyo. I-Cloud4Y ithathe isigqibo sokuyilungisa le nto ishiyiweyo.

Makhe sithethe ngezona projekthi zinika umdla eziphunyezwayo eAfrika.

I-DeepMind ilandelela imihlambi yeSerengeti

Kwiminyaka eyi-10 edlulileyo, izazi zebhayoloji, i-ecologists kunye namavolontiya olondolozo kwiprogram ye-Serengeti Lion Research baye baqokelela kwaye bahlalutya idatha evela kumakhulu eekhamera zentsimi eziseSerengeti National Park (eTanzania). Oku kuyimfuneko ukufunda indlela yokuziphatha kweentlobo ezithile zezilwanyana ezibukho bazo busesichengeni. Amavolontiya achithe unyaka wonke elungisa ulwazi, efunda amanani abantu, iintshukumo kunye nezinye iimpawu zemisebenzi yezilwanyana. I-AI DeepMind sele isenza lo msebenzi kwiinyanga ze-9.

I-DeepMind yinkampani yaseBhritane ephuhlisa itekhnoloji yobukrelekrele bokwenziwa. Ngo-2014, yathengwa yi-Alphabet. Ukusebenzisa idataset Umfanekiso weSerengeti ukuqeqesha imodeli yobukrelekrele bokwenziwa, iqela lophando lifumene iziphumo ezigqwesileyo: AI DeepMind iyakwazi ukubona ngokuzenzekelayo, ukuchonga nokubala izilwanyana zaseAfrika kwimifanekiso, ukwenza umsebenzi wayo kwiinyanga ezi-3 ngokukhawuleza. Abasebenzi be-DeepMind bachaza ukuba kutheni oku kubalulekile:

“I-Serengeti yenye yeendawo ezisaseleyo emhlabeni apho kukho izidalwa ezikhulu ezanyisayo... Njengoko ungenelelo lwabantu olujikeleze le paki lusiba luqilima, ezi zilwanyana zinyanzeleka ukuba zitshintshe indlela eziziphethe ngayo ukuze ziphile. Ukwanda kwezolimo, ukuzingela ngokungekho mthethweni kunye nokungahambi kakuhle kwemozulu kubangela utshintsho kwindlela yokuziphatha kwezilwanyana kunye nokutshintsha kwabemi, kodwa olu tshintsho lwenzekile kwimilinganiselo yendawo kunye neyexeshana ekunzima ukuyijonga kusetyenziswa iindlela zophando eziqhelekileyo. "

Kutheni le nto ubukrelekrele bokwenziwa busebenza ngokufanelekileyo kunobukrelekrele bebhayoloji? Kukho izizathu ezininzi zoku.

• Ezinye iifoto zibandakanyiwe. Ukusukela ukufakela, iikhamera zentsimi zibambe amakhulu ezigidi zemifanekiso. Ayizizo zonke ekulula ukuzibona, ngoko ke amavolontiya kufuneka azichonge ngesandla iintlobo esebenzisa isixhobo sewebhu esibizwa ngokuba yiZooniverse. Ngoku kukho iindidi ezingama-50 ezahlukeneyo kwiziko ledatha, kodwa ixesha elininzi lichithwe ekusetyenzweni kwedatha. Ngenxa yoko, ayizizo zonke iifoto ezisetyenziswayo emsebenzini.
• Ukuqatshelwa kweentlobo ezikhawulezayo. Inkampani ithi inkqubo yayo yokuqeqeshwa kwangaphambili, eya kuthunyelwa ngokukhawuleza kwintsimi, iyakwazi ukwenza ngokuhambelanayo (okanye nokuba ngcono kune) izichazisi zoluntu ekukhumbuleni nasekuqapheleni iintlobo ezingaphezu kwekhulu zezilwanyana ezifumaneka kummandla.
• Izixhobo ezitshiphu. I-AI DeepMind iyakwazi ukusebenza ngokufanelekileyo kwi-hardware ethobekileyo enokufikelela kwi-Intanethi okungathembekanga, okuyinyani ngokukodwa kwilizwekazi laseAfrika, apho iikhomputha ezinamandla kunye nokufikelela kwi-Intanethi ngokukhawuleza kunokonakalisa izilwanyana zasendle kwaye kubize kakhulu ukuzisebenzisa. Ukhuseleko lwebhayoloji kunye nogcino lweendleko ziinzuzo ezibalulekileyo ze-AI kumatsha-ntliziyo okusingqongileyo.

Inkqubo yokufunda yomatshini ye-DeepMind ilindeleke ukuba ingakwazi nje ukulandelela ukuziphatha kwabantu kunye nokusabalalisa ngokubanzi, kodwa iphinde ibonelele ngedatha ngokukhawuleza ukuvumela abalondolozi bezolondolozo ukuba baphendule ngokukhawuleza kwiinguqu zexesha elifutshane ekuziphatheni kwezilwanyana zaseSerengeti.

I-Microsoft ilandelela iindlovu

Ukulunga, siyaqaphela ukuba i-DeepMind ayisiyiyo kuphela inkampani ejongene nokusindisa abantu ababuthathaka bezilwanyana zasendle. Ke, iMicrosoft yavela eSanta Cruz ngokuqalwa kwayo Iimetriki zoLondolozo, esebenzisa i-AI ukulandelela iindlovu ze-savannah zase-Afrika.

Ukuqala, inxalenye yeProjekthi yokuMamela iiNdlovu, ngoncedo oluvela kwibhubhoratri kwiYunivesithi yaseCornell, iphuhlise inkqubo ekwazi ukuqokelela kunye nokuhlalutya idatha evela kwi-acoustic sensors esasazwe kuyo yonke i-Nouabale-Ndoki National Park kunye nemimandla yamahlathi ajikelezileyo kwiRiphabhliki yaseCongo. I-Artificial intelligence iyaliqonda ilizwi leendlovu kushicilelo-izandi zokuduma eziphantsi-frequency ezizisebenzisayo ukunxibelelana enye kwenye, kwaye ifumana ulwazi malunga nobukhulu bomhlambi kunye necala lentshukumo yawo. Ngokutsho kwe-CEO ye-Conservation Metrics u-Matthew McKone, ubukrelekrele bokwenziwa bunokuchonga ngokuchanekileyo isilwanyana ngasinye esingenakubonwa emoyeni.

Okubangela umdla kukuba, le projekthi ibe nesiphumo sophuhliso lwe-algorithm yokufunda ngomatshini oqeqeshwe kwi-Snapshot Serengeti enokuchonga, ukuchaza nokubala. izilwanyana zasendle ngokuchaneka kwama-96,6%.

I-TrailGuard Resolve ilumkisa malunga nabazingeli abangekho mthethweni

Ikhamera ye-Intel-powered smart isebenzisa i-AI ukukhusela izilwanyana zasendle zase-Afrika ezisengozini kubazingeli. Eyona nto ibalaseleyo yale nkqubo kukuba ilumkisa ngamalinge okubulala izilwanyana ngokungekho mthethweni kwangaphambili.

Iikhamera ezibekwe kuyo yonke ipaki zisebenzisa iprosesa yembono yekhompyuter ye-Intel (i-Movidius Myriad 2) ekwazi ukubona izilwanyana, abantu kunye nezithuthi ngexesha lokwenyani, ivumela abagcini beepaki ukuba babambe abazingeli ngaphambi kokuba benze nantoni na engalunganga.

Itekhnoloji entsha ethi i-Resolve iza nezithembiso eziya kusebenza ngakumbi kunezinzwa zokubona eziqhelekileyo. Iikhamera ezichasene nokuzingela zithumela izilumkiso nanini na xa zibona ukunyakaza, okukhokelela kwiialamu ezininzi zobuxoki kunye nokunciphisa ubomi bebhetri kwiiveki ezine. Ikhamera yeTrailGuard isebenzisa kuphela intshukumo ukuvusa ikhamera kwaye ithumela kuphela izilumkiso xa ibona abantu kwisakhelo. Oku kuthetha ukuba kuya kuba mbalwa kakhulu iimpawu zobuxoki.

Ukongeza, ikhamera yokuSombulula ayisebenzisi mandla kwimowudi yokulinda kwaye inokuhlala ukuya kuthi ga kunyaka onesiqingatha ngaphandle kokutshaja kwakhona. Ngamanye amazwi, abasebenzi basepakini abayi kubeka emngciphekweni ukhuseleko lwabo rhoqo njengangaphambili. Ikhamera ngokwayo ilingana nepensile, nto leyo eyenza kube lula ukufunyanwa ngabazingeli abangekho mthethweni.

Yintoni enye onokuyifunda kwiblogi? Cloud4Y

→ vGPU - ayinakuhoywa
→ Ubukrelekrele bebhiya-i-AI iza nebhiya
→ Iindlela ezi-4 zokugcina kwii-backups zamafu
→ Ii-Kubernetes Distros ezi-5 ezilungileyo
→ Iirobhothi kunye namaqunube: indlela i-AI eyandisa ngayo imveliso yasendle

Bhalisela yethu yocingo-ijelo ukuze ungaphoswa linqaku elilandelayo! Asibhali ngaphezu kwesibini ngeveki kwaye kuphela kwishishini.

umthombo: www.habr.com