I-Singapore ithuthukise i-AI eyisebenzisayo ukuthola abantu ababhemayo abakhanyisa ezindaweni eziningi lapho lo mkhuba ungavunyelwe kuso sonke isiqhingi, ukuze usize abomthetho bendawo bavimbele izephula-mthetho ngokuphumelelayo.
I-AI ibizwa nge-Balefire, futhi muva nje wachaza nguPye Sone Kyaw – unjiniyela we-AI esikhungweni sokuguqula idijithali saseSingapore i-GovTech – isivele ifinyelelwe kunguqulo 3.0.
“Inhloso enkulu ye-Balefire … wukusiza i-NEA [i-National Environment Agency] ekutholeni ababhemayo ezindaweni okungavunyelwe ukubhema,” ebhala. I-NEA iyasiza izinhlu lezo zindawo ezingavunyelwe: izindawo eziningi zasendlini, amapaki, izikhungo zemfundo, amachibi okubhukuda, ngisho namabhuloho ahamba ngezinyawo. Izinhlawulo zika-S$200 ($148) zingakhokhiswa ukubhema endaweni engafanele, futhi ukugwetshwa ngecala kungase kubangele inhlawulo ephindwe kahlanu kuleyo samba.
Izinguqulo zangaphambilini ze-Balefire zazibhekwa njengobufakazi bomqondo wamademo. Inguqulo engu-3.0 ibhekwa “njengomshayeli wendiza owandisiwe” osebenza ezindaweni ezingu-20.
UKyaw ukhale ngokuthi ukubona ugwayi akulula – bancane futhi kulula ukuphambanisa nezinye izinto. Wathi “izitshani, imiphetho yocingo ecwebezelayo, iminwe ebekwe ngezindlela ezithile, ngisho nezinhlobo ezithile zokudla” njengezinto amasistimu okubona amakhompyutha ancike kumakhamera angaphandle angazibona njengenduku yomdlavuza.
Wazama ukuthola intuthu noma ithiphu elivuthayo likagwayi, kodwa leyo mizamo yaphela ngoba yaveza amaphutha amaningi kakhulu. Kanjalo “nokubheka umuntu ewonke, njengokulinganisa kokuma.”
Lokho kwehluleka kuholele ekutheni u-Kyaw aphethe ngokuthi “imodeli yokutholwa kokuphela ukuya ekupheleni ayinakwenzeka, ikakhulukazi kumongo we-AI onqenqemeni kanye nemikhawulo yayo yekhompiyutha engokwemvelo kanye nosayizi wamamodeli amancane uma kuqhathaniswa, okuhambisana nesidingo sokutholwa ngokushesha.”
Wabheka amasistimu angekho eshalofini angase athuthuke ku-Balefire, kodwa akatholanga noma yiziphi ezihlangabezana nesidingo se-NEA sesistimu ekwazi ukuhlonza ababhemayo abaningi ngangokunokwenzeka kuso sonke isikhathi senkambu yokubuka yekhamera nokwenza cishe lokho. khona manjalo.
Ngakho-ke i-GovTech yakha ipayipi layo lokucubungula ngokwezifiso elabhalwa nguKyaw. Kubandakanya izinyathelo ezinhlanu ezilandelayo:
- Ukutholwa Kwekhanda Nokucubungula: Ipayipi liqala ngokuthi amafreyimu ekhamera afakwe kumtshina wekhanda, okhomba izixhumanisi zawo wonke amakhanda ngaphakathi kohlaka.
- Ukuhlunga Okusekelwe ku-Heuristic: Ngemuva kokutholwa, lawa makhanda athola uchungechunge lwezihlungi ze-heuristic eziklanyelwe ukuqeda amakhanda angaba yiphutha. Lezi zihlungi ziwumkhiqizo wokufunda okunqwabelene nokuhlaziywa okuningiliziwe kwedatha yokusebenzisa.
- Ukulandelela ikhanda: I-tracker yento ibe ilandela amakhanda atholiwe kuwo wonke amafreyimu alandelanayo, iwaxhumanise namakhanda atholwe ngaphambilini lapho kunokwenzeka khona. Lokhu kuqinisekisa ukuthi, kubantu ababhemayo abahlonziwe, izexwayiso eziphindaphindiwe aziqalwa isikhathi ngasinye lapho zibonwa ohlakeni olusha.
- Ukuhlukaniswa Kwentuthu/Intuthu: Amakhanda angazange achazwe ngaphambilini njengawabantu ababhemayo abe esecutshungulwa kusetshenziswa i-binary head classifier. Lesi sigaba sinquma ukuthi umuntu uyabhema noma cha.
- Ukuphinda Ukuhlonza Imojuli: Uma ohlukanisa ngezigaba ebonisa umsebenzi wokubhema, imojuli yokuphinda ihlonze izama ukufanisa obhemayo otholiwe ngokumelene nohlu lokubuka lwababhemayo bakamuva. Uma kungekho ukuhlonzwa kabusha, isexwayiso siyaqaliswa. Uhlu lokubuka lubuyekezwa ngokubukeka kwakamuva komuntu obhemayo nolunye ulwazi olufanele.
Inguqulo engu-3.0 isebenzisa amamodeli amaningi adweba ezithombeni ezishuthwe kusukela ekuphindaphindweni kwamanje nokudlule kwe-Balefire.
"Kalula nje, sisebenzise amamodeli ethu akhona ukusichasisela idatha entsha futhi salungisa noma yimaphi amaphutha aleyo nqubo," kubhala u-Kyaw. “Singeze ngokuphindaphindiwe kumaphrofayili athile ezithombe amamodeli akhona ayethambekele kuwo amaphutha, njengabantu abagqoke izigqoko zokuzivikela, noma abantu abadlayo noma abaphuzayo. Lokhu kusize ekuthuthukiseni ukusebenza kwamamodeli kakhulu phakathi nephrojekthi. "
Lolu hlelo olusha lwethenjwa ukuthi ngeke lugcine ngokuthola abantu ababhemayo abaningi kuphela, kodwa futhi lugweme imibono engamanga - “ukusiza i-NEA ekwenzeni ngcono ukwabiwa kwamaphoyisa kulezi zindawo ezithandwa kakhulu.”
Ngamanye amazwi, i-Balefire ihlose ukuqinisekisa ukuthi uma i-NEA ihlasela ababhemayo, imizamo yayo ayibi ngumlotha. ®
- I-SEO Powered Content & PR Distribution. Khuliswa Namuhla.
- I-PlatoData.Network Vertical Generative Ai. Zinike Amandla. Finyelela Lapha.
- I-PlatoAiStream. I-Web3 Intelligence. Ulwazi Lukhulisiwe. Finyelela Lapha.
- I-PlatoESG. Ikhabhoni, I-CleanTech, Amandla, Environment, Ilanga, Ukuphathwa Kwemfucuza. Finyelela Lapha.
- I-PlatoHealth. I-Biotech kanye ne-Clinical Trials Intelligence. Finyelela Lapha.
- Source: https://go.theregister.com/feed/www.theregister.com/2024/03/28/singapore_balefire_smoking_detection_ai/
