I-AI isiza ukufunda izilwane zase-Afrika

Kusuka kunoma yiliphi iketela likagesi elixhunywe ku-inthanethi, ungezwa ngendlela i-AI ehlula ngayo abasubathi be-cyber, inikeza amathuba amasha kubuchwepheshe obudala, futhi idonsa amakati ngokusekelwe kumdwebo wakho. Kodwa abavamile ukukhuluma ngeqiniso lokuthi ubuhlakani bemishini buyakwazi nokunakekela imvelo. I-Cloud4Y inqume ukulungisa lokhu kushiywa.

Ake sikhulume ngamaphrojekthi athakazelisa kakhulu asetshenziswa e-Afrika.

I-DeepMind ilandelela imihlambi ye-Serengeti

Kule minyaka eyi-10 edlule, izazi zesayensi yezinto eziphilayo, izazi zemvelo kanye namavolontiya okulondoloza imvelo ohlelweni lwe-Serengeti Lion Research bebeqoqa futhi bahlaziya idatha evela kumakhulu amakhamera atholakala eSerengeti National Park (Tanzania). Lokhu kuyadingeka ukuze utadishe ukuziphatha kwezinhlobo ezithile zezilwane okusongelwa ubukhona bazo. Amavolontiya achithe unyaka wonke ecubungula ulwazi, efunda izibalo zabantu, ukunyakaza nezinye izimpawu zemisebenzi yezilwane. I-AI DeepMind isivele yenza lo msebenzi ezinyangeni eziyisi-9.

I-DeepMind yinkampani yaseBrithani eyakha ubuchwepheshe bobuhlakani bokwenziwa. Ngo-2014, yathengwa nge-Alphabet. Ukusebenzisa idathasethi Isithombe se-Serengeti ukuqeqesha imodeli yobuhlakani bokwenziwa, ithimba labacwaningi lithole imiphumela emihle kakhulu: I-AI DeepMind ikwazi ukubona ngokuzenzakalelayo, ibone futhi ibale izilwane zase-Afrika ezithombeni, okwenza umsebenzi wayo ube nezinyanga ezi-3 ngokushesha. Abasebenzi be-DeepMind bachaza ukuthi kungani lokhu kubalulekile:

“I-Serengeti ingenye yezindawo ezisasele emhlabeni ezinomphakathi ongaphelele wezilwane ezincelisayo ezinkulu... Njengoba ukungenela kwabantu epaki kuya kushuba, lezi zilwane ziphoqeleka ukuba zishintshe ukuziphatha kwazo ukuze ziphile. Ukwenyuka kwezolimo, ukuzingela ngokungemthetho kanye nezimo zezulu kubangela izinguquko ekuziphatheni kwezilwane kanye nokuguquguquka kwenani labantu, kodwa lezi zinguquko zenzeke ezilinganisweni zendawo nezesikhashana okunzima ukuziqapha kusetshenziswa izindlela zocwaningo ezivamile.”

Kungani ubuhlakani bokwenziwa busebenza ngempumelelo kunobuhlakani bebhayoloji? Kunezizathu eziningana zalokhu.

• Izithombe ezengeziwe zifakiwe. Kusukela ekufakweni, amakhamera asensimini athwebule izithombe eziyizigidi ezingamakhulu ambalwa. Akubona bonke okulula ukuzibona, ngakho-ke amavolontiya kufanele azibonele mathupha lezi zinhlobo zisebenzisa ithuluzi lewebhu elibizwa ngokuthi i-Zooniverse. Njengamanje kunezinhlobo ezingu-50 ezihlukene kusizindalwazi, kodwa isikhathi esiningi sichithwa kucubungula idatha. Ngenxa yalokho, akuzona zonke izithombe ezisetshenziswa emsebenzini.
• Ukuqashelwa kwezinhlobo ezisheshayo. Le nkampani ithi uhlelo lwayo oluqeqeshwe kusengaphambili, oluzosetshenziswa maduze nje kulo mkhakha, luyakwazi ukusebenza ngokulingana (noma okungcono nakakhulu kuna) izichasiselo ezingabantu ekukhumbuleni nasekuqapheleni izinhlobo zezilwane ezingaphezu kwekhulu ezitholakala endaweni.
• Imishini eshibhile. I-AI DeepMind iyakwazi ukusebenza kahle ku-Hardware enesizotha enokufinyelela ku-inthanethi okungathembekile, okuyiqiniso ikakhulukazi ezwenikazi lase-Afrika, lapho amakhompyutha anamandla nokufinyelela ku-inthanethi okusheshayo kungalimaza izilwane zasendle futhi kubize kakhulu ukukusebenzisa. I-Biosecurity nokonga izindleko kuyizinzuzo ezibalulekile ze-AI kubashisekeli bemvelo.

Isistimu yokufunda yomshini ye-DeepMind kulindeleke ukuthi ingagcini nje ngokukwazi ukulandelela ukuziphatha nokusabalalisa kwabantu ngokuningiliziwe, kodwa futhi inikeze idatha ngokushesha ngokwanele ukuze ivumele abalondolozi bemvelo ukuthi baphendule ngokushesha ezinguqukweni zesikhathi esifushane ekuziphatheni kwezilwane ze-Serengeti.

I-Microsoft ilandelela izindlovu

Uma sikhuluma iqiniso, siyaqaphela ukuthi i-DeepMind akuyona ukuphela kwenkampani ethintekayo ngokusindisa isibalo esintekenteke sezilwane zasendle. Ngakho-ke, iMicrosoft yavela eSanta Cruz ngokuqala kwayo I-Conservation Metrics, esebenzisa i-AI ukulandelela izindlovu zase-African savannah.

Ukuqala, okuyingxenye yoHlelo Lokulalela Izindlovu, ngosizo oluvela kulabhorethri yaseNyuvesi yase-Cornell, selusungule uhlelo olukwazi ukuqoqa nokuhlaziya idatha evela kuzinzwa ze-acoustic ezisabalele kuyo yonke i-Nouabale-Ndoki National Park nasezindaweni ezizungezile zamahlathi e-Republic of Congo. I-Artificial intelligence iyaliqaphela izwi lezindlovu ekurekhodweni - imisindo edumayo ephansi eziyisebenzisela ukuxhumana zodwa, futhi ithola ulwazi mayelana nobukhulu bomhlambi kanye nesiqondiso sokuhamba kwawo. Ngokusho kwesikhulu se-Conservation Metrics u-Matthew McKone, ubuhlakani bokwenziwa bungakwazi ukubona ngokunembile isilwane ngasinye esingabonakali emoyeni.

Kuyathakazelisa ukuthi le phrojekthi ibe nomphumela wokwakhiwa kwe-algorithm yokufunda komshini eqeqeshwe ku-Snapshot Serengeti ekwazi ukuhlonza, ukuchaza nokubala. izilwane zasendle ngokunemba okungu-96,6%.

I-TrailGuard Resolve ixwayisa ngabazingeli abangekho emthethweni

Ikhamera ehlakaniphile ene-Intel isebenzisa i-AI ukuvikela izilwane zasendle zase-Afrika ezisengozini kubazingeli. Isici salolu hlelo ukuthi luxwayisa ngemizamo yokubulala izilwane ngokungemthetho kusenesikhathi.

Amakhamera atholakala kulo lonke elasepaki asebenzisa i-Intel computer vision processor (Movidius Myriad 2) ekwazi ukubona izilwane, abantu nezimoto ngesikhathi sangempela, evumela abaqaphi bepaki ukuthi babambe abazingeli ngaphambi kokuba benze noma yini engalungile.

Ubuchwepheshe obusha okuthi Ukuxazulula buqhamuke nezithembiso ezizosebenza kangcono kunezinzwa ezijwayelekile zokubona. Amakhamera amelene nokuzingelwa ngokungemthetho athumela izexwayiso noma nini lapho ebona ukunyakaza, okuholela kuma-alamu amaningi angamanga futhi akhawulele ukuphila kwebhethri kumaviki amane. Ikhamera ye-TrailGuard isebenzisa kuphela ukunyakaza ukuze ivuse ikhamera futhi ithumela kuphela izexwayiso lapho ibona abantu kuhlaka. Lokhu kusho ukuthi kuzoba namaphozithivu ambalwa kakhulu.

Ngaphezu kwalokho, ikhamera yokuxazulula ayisebenzisi cishe amandla kumodi yokulinda futhi ingahlala isikhathi esingafika unyaka nesigamu ngaphandle kokushajwa kabusha. Ngamanye amazwi, abasebenzi basepaki ngeke kudingeke babeke ukuphepha kwabo engozini kaningi njengangaphambili. Ikhamera ngokwayo ilingana nepensela, okwenza kube mancane amathuba okuthi itholwe abazingeli.

Yini enye ongayifunda kubhulogi? Cloud4Y

→ I-vGPU - ayikwazi ukuzitshwa
→ Ubuhlakani bobhiya - i-AI iza nobhiya
→ Izindlela ezi-4 zokulondoloza kuma-backups wamafu
→ Top 5 Kubernetes ukusatshalaliswa
→ Amarobhothi namasitrobheni: ukuthi i-AI ithuthukisa kanjani isivuno sezitshalo

Bhalisela yethu yocingo-isiteshi, ukuze ungaphuthelwa yisihloko esilandelayo! Asibhali ngaphezu kokuphindwe kabili ngesonto futhi ngebhizinisi kuphela.

Source: www.habr.com