Hiki iā mākou ke hana i kahi honua me ka maʻi ʻole?

Hiki ke loaʻa i kahi honua maʻi ʻole? ʻO ka maʻi ka huaʻōlelo i ka hapanui (inā ʻaʻole ʻo nā mea a pau) ʻoluʻolu ʻole ka lohe ʻana inā loaʻa iā lākou a i ʻole kekahi mea a lākou i ʻike ai. ʻO ka pōmaikaʻi, Max Welling, he polopeka o ka mīkini aʻo ma ke Kulanui o Amsterdam a me ka lālā o ka Canadian Institute of Advanced noiʻi, a me kāna hui o nā ʻoihana i hana i kahi ʻōnaehana ʻikepili no ka ʻike ʻana i nā maʻi no nā maʻi. ʻO ka ʻoiaʻiʻo leʻaleʻa: Ke alakaʻi nei ʻo ia i ka AMLAB (Amsterdam Machine Learning LAB) a me ke alakaʻi pū ʻana iā QUVA Lab (Qualcomm-UvA Lab). Ma ʻaneʻi mākou e ʻike ai i ka hana ʻana o kēia kanaka kupanaha a me kāna hui o nā ʻoihana (Cynthia Dwork, Geoffrey Hinton a me Judea Pearl) i kekahi mau hiʻohiʻona kupaianaha e hoʻopau i ka honua mai ka maʻi.

ʻO nā hopohopo o Max Welling

ʻO kekahi o nā mea ʻoiaʻiʻo a Welling i hōʻike ai i kāna kamaʻilio TEDx e lawe mai i ka manaʻo i ka ʻoiaʻiʻo aia kekahi mau manawa e poina ai ke kauka i kekahi mea i ka wā o ka maʻi maʻi. No ka laʻana, ʻōlelo ʻo ia "ʻaʻohe hapa o nā kaʻina hana lapaʻau e noi i nā hōʻike ʻepekema." Hana ʻia kēlā hōʻailona ma o kā lākou hana ponoʻī a me ka ʻike i loaʻa i ke kula, ʻoiai ʻo Max e ʻōlelo nei aia kekahi ʻano o ka analytical prognosis no nā maʻi ʻē aʻe. Hoʻomau ʻo ia i ka wehewehe ʻana e hiki ke kuhi hewa ʻia kekahi mau maʻi a hoʻi hou i ka haukapila, kahi āna i kuhikuhi ai he 8 mau manawa e make ai lākou. ʻO ka mea hoihoi loa, ʻo ia ka pilikia i loaʻa mau. He mea maʻalahi ke kumu e like me ka hana hewa ʻana i hiki ke lilo i ke ola o kekahi a i ʻole kekahi poʻe. ʻAʻole ʻo ia wale nō, e like me kā Welling i ʻōlelo ai, aia he 230 miliona mau hana lapaʻau i kēlā me kēia makahiki e kūʻai ana i ka hapalua trillion kālā. E like me kēlā me kēia ʻoihana e hoʻāʻo nei e hāʻawi i kahi lawelawe e kōkua i nā poʻe ʻē aʻe, koi ke kālā; Eia kekahi, ʻo ia hoʻi, pono e hoʻolohe nā halemai a me ka poʻe i mālama i ke kālā i nā kikowaena olakino i nā mea hou e hoʻāʻo nei e hoʻonui i ka ʻoihana i kahi ala maikaʻi aʻe. Eia nō naʻe, ʻoi aku ka maikaʻi o ka mālama kālā.

Mālama i ka pilikino

Ua ʻōlelo ʻo Welling ua hana ʻo ia a me kāna hui i 3 holomua. ʻO kekahi o ia mea he kamepiula hiki ke mālama i ka pilikino i loko o kahi haukapila; Eia kekahi, hiki i nā lolouila ke kālailai i ka nui o nā ʻikepili e hoʻomaikaʻi hou aku i ka maʻi maʻi no nā maʻi maʻi. Kapa ʻia kēia polokalamu Mīkini aʻo. ʻO ka mea nui, hoʻouna ke kamepiula i kahi nīnau i ka waihona o ka haukapila, e pane ana i ka nīnau a laila e hoʻololi ka mea aʻo mīkini i ka pane ma ka "hoʻohui ʻana i kahi walaʻau iā ia." No ka ʻike hou aku e ʻoluʻolu Bus maanei (Ua wehewehe ʻo Max Welling iā ia ma waena o nā minuke 5:20 – 6:06). I nā huaʻōlelo ʻē aʻe, e like me ka wehewehe ʻana o Max, makemake ke kamepiula e "ʻoi aku ka maikaʻi" ma o ka maʻi ʻana a "kūkulu i kahi kumu hoʻohālike maikaʻi aʻe". Mahalo kēia mau mea a pau Cynthia Dwork, he kanaka ʻepekema kaulana mai Microsoft Research. Hoʻopili ʻo ia i ka mālama ʻana i ka pilikino ma muli o ke kumu makemakika. No nā mea hou aku iā ia a me kāna i hana ai, Bus maanei. I ka pōkole, ʻaʻole i hōʻike wale kēia holomua mua e makemake ana ʻo Max e mahalo i ka ʻike pilikino o ka poʻe maʻi akā makemake nō hoʻi e hoʻolako i nā halemai i kahi kumu paʻa no ka maʻi.

Hoʻopiha hohonu

Ua hōʻike ʻia ka holomua lua e ʻO Geoffrey Hinton. ʻO Yann Lecun, Yoshua Bengio a me Geoffrey Ua wehewehe ʻo ia: "ʻIke ʻia ke aʻo hohonu i ke ʻano paʻakikī i nā pūʻulu ʻikepili nui ma o ka hoʻohana ʻana i ka algorithm backpropagation e hōʻike i ke ʻano o ka mīkini e hoʻololi ai i kāna mau ʻāpana kūloko i hoʻohana ʻia e helu i ka hōʻike i kēlā me kēia papa mai ka hōʻike ʻana i ka papa mua." Ma nā huaʻōlelo layman, kōkua ia i kahi mīkini e hoʻomaopopo maikaʻi iā ia iho ma o kāna mau papa paʻakikī ma o kāna mau ʻāpana hohonu loa (no ka ʻike hou aku e heluhelu i ke koena o ka loiloi i kākau ʻia e nā keonimana ʻekolu).

Kauality vs Correlation

ʻO ke kolu a me ka hopena hope loa, ʻo ia ka manaʻo hui e hoʻokaʻawale hou i ka kumu kumu mai ka hoʻopili ʻana. Manaʻo ʻo Max e hiki i nā mea hana a Judea Pearl ke kōkua i ka hoʻokaʻawale ʻana i kēia mau manaʻo ʻelua a e hoʻonohonoho pono. ʻO ka mea nui, ʻo ka hana a Iudea ke kōkua i ka hāʻawi ʻana i nā ʻikepili hou aʻe i hiki ke hana inā hoʻololi ʻia nā faila maʻi i loko o kahi waihona. He paʻakikī ka hana a Pearl no laila inā makemake ʻoe e hoʻomaopopo hou i kāna mau "mea hana". Bus maanei.

Makemake o Max

Ua hōʻuluʻulu ʻia ʻo Welling ma ka hope o kāna Kūkākūkā TEDX makemake ʻo ia e mālama i ka pilikino ma o ka mea aʻo mīkini. ʻO ka lua, e hoʻopili i nā keiki liʻiliʻi a me nā ʻepekema i mea e hoʻomaikaʻi hou ai i ka maʻi maʻi e mālama i ke kālā a me nā ola. ʻO ka hope, makemake ʻo ia e hoʻololi i ka mālama olakino ma o ka lawelawe maikaʻi ʻana i nā haukapila, nā kauka a me nā mea maʻi ma o ka ʻenehana e hiki ke kōkua i ka hōʻemi ʻana i nā kipa haukapila a hoʻohana pono i ke kālā. He hiʻohiʻona nani kēia no ka mālama ola kino no ka mea ʻaʻole makemake wale ʻo ia e mahalo i ka ʻoihana lapaʻau, makemake pū ʻo ia e kōkua i ka hoʻopakele ʻana i nā ola me ka noʻonoʻo ʻana i nā kumukūʻai o nā halemai.