Hai Habr!
Muna so muyi magana game da irin wannan hanyar da ba a saba ba don samun kuɗi ga matasa kamar shiga cikin hackathons. Wannan duka yana da fa'ida ta kuɗi kuma yana ba ku damar aiwatar da ilimin da aka samu a makaranta da kuma ta hanyar karanta littattafai masu wayo.
Misali mai sauƙi shine na bara na Artificial Intelligence Academy hackathon don yaran makaranta. Mahalarta taron sun yi hasashen sakamakon wasan Dota 2. Wanda ya lashe gasar shi ne Alexander Mamaev, dan aji goma daga Chelyabinsk. Algorithm ɗin sa ya fi dacewa ya ƙayyade ƙungiyar da ta yi nasara. Godiya ga wannan, Alexander samu wani gagarumin kudi kyauta - 100 rubles.
Ta yaya Alexander Mamaev ya yi amfani da kuɗin kyauta, abin da ilimin da dalibi ya rasa don yin aiki tare da ML, da kuma wane shugabanci a fagen AI ya ɗauki mafi ban sha'awa - dalibi ya fada a cikin wata hira.
- Faɗa mana game da kanku, ta yaya kuka zama masu sha'awar AI? Shin yana da wuya a shiga cikin batun?
— Ina ɗan shekara 17, ina gama makaranta a wannan shekara, kuma kwanan nan na ƙaura daga Chelyabinsk zuwa Dolgoprudny, da ke kusa da Moscow. Ina karatu a Kapitsa Physics and Technology Lyceum, wannan shine ɗayan mafi kyawun makarantu a yankin Moscow. Zan iya yin hayan gida, amma ina zaune a makarantar kwana a makarantar, yana da kyau kuma mafi sauƙi don sadarwa tare da mutane daga lyceum.
Lokaci na farko da na ji game da AI da ML yana yiwuwa a cikin 2016, lokacin da Prisma ta bayyana. Daga nan ina aji 8 kuma ina yin programming na olympiad, na halarci wasu wasannin Olympics na gano cewa muna taron ML a cikin birni. Ina sha'awar gano shi, fahimtar yadda yake aiki, kuma na fara zuwa can. A can na fara koyon abubuwan yau da kullun, sannan na fara karanta shi a Intanet, a cikin kwasa-kwasan darussa daban-daban.
Da farko, akwai kawai hanya daga Konstantin Vorontsov a cikin Rashanci, kuma hanyar koyarwa ta kasance mai tsauri: ya ƙunshi sharuddan da yawa, kuma a cikin kwatancin ya kasance da yawa. Ga dalibi na takwas wannan yana da matukar wahala, amma yanzu, daidai saboda na shiga irin wannan makaranta a farkon, sharuɗɗan ba su da wahala a gare ni a aikace a cikin matsalolin gaske.
- Nawa ilimin lissafi kuke buƙatar sani don aiki tare da AI? Shin akwai isasshen ilimi daga tsarin karatun makaranta?
- A hanyoyi da yawa, ML ya dogara ne akan ainihin ra'ayoyin makaranta a maki 10-11, algebra na asali na layi da bambanci. Idan muna magana ne game da samarwa, game da matsalolin fasaha, to ta hanyoyi da yawa ba a buƙatar ilimin lissafi, yawancin matsaloli ana magance su kawai ta hanyar gwaji da kuskure. Amma idan muka yi magana game da bincike, lokacin da aka kirkiro sabbin fasahohi, to babu inda babu lissafi. Ana buƙatar lissafi a matakin asali, aƙalla don sanin yadda ake amfani da matrix ko, in mun gwada da magana, ƙididdige abubuwan da aka samo asali. Babu tserewa ilimin lissafi a nan.
- A ra'ayin ku, kowane ɗalibi mai tunani-bincike zai iya magance matsalolin ML?
- Da. Idan mutum ya san abin da ke cikin zuciyar ML, idan ya san yadda aka tsara bayanai kuma ya fahimci dabaru na asali ko hacks, ba zai buƙaci lissafi ba, saboda yawancin kayan aikin da aka riga aka rubuta ta wasu mutane. Duk yana zuwa nemo alamu. Amma komai, ba shakka, ya dogara da aikin.
- Menene abu mafi wahala wajen magance matsalolin ML da lokuta?
- Kowane sabon aiki sabon abu ne. Da a ce matsalar ta riga ta wanzu a cikin tsari iri ɗaya, da ba sai an warware ta ba. Babu algorithm na duniya. Akwai ɗimbin jama'a waɗanda ke horar da dabarun warware matsalolinsu, suna ba da labarin yadda suka magance matsalolin, da bayyana labaran nasarorin da suka samu. Kuma yana da ban sha'awa sosai don bin dabaru, ra'ayoyinsu.
- Wadanne matsaloli da matsaloli kuka fi sha'awar warwarewa?
- Na ƙware a ilimin harshe na lissafi, Ina sha'awar rubutu, matsalolin rarrabawa, chatbots, da sauransu.
- Shin kuna yawan shiga AI hackathons?
- Hackathons, a gaskiya, tsarin tsarin Olympiad ne daban-daban. Olympiad yana da rufaffiyar matsalolin, tare da sanannun amsoshi waɗanda dole ne ɗan takara ya yi tsammani. Amma akwai mutanen da ba su da kyau a rufaffiyar ayyuka, amma suna yayyaga kowa a bude. Don haka kuna iya gwada ilimin ku ta hanyoyi daban-daban. A cikin matsalolin da aka buɗe, ana ƙirƙira fasahohi a wasu lokuta daga karce, samfuran suna haɓaka da sauri, har ma masu shiryawa sau da yawa ba su san amsar daidai ba. Sau da yawa muna shiga cikin hackathons, kuma ta wannan za mu iya samun kuɗi. Wannan yana da ban sha'awa.
- Nawa za ku iya samu daga wannan? Yaya kuke kashe kuɗin kyautar ku?
- Ni da abokina mun shiga cikin VKontakte hackathon, inda muka yi aikace-aikacen neman zane-zane a cikin Hermitage. An nuna saitin emoticons da emoticons akan allon wayar, ya zama dole a nemo hoto ta amfani da wannan saitin, an nuna wayar akan hoton, an gane ta ta amfani da hanyoyin sadarwa na neural kuma, idan amsar ta kasance daidai, an ba da maki. Mun yi farin ciki da sha'awar cewa mun sami damar ƙirƙirar aikace-aikacen da ke ba mu damar gane zane akan na'urar hannu. Mun kasance a farkon wuri, amma saboda ƙa'idar doka mun rasa kyautar 500 dubu rubles. Abin kunya ne, amma wannan ba shine babban abin ba.
Bugu da kari, ya halarci gasar Sberbank Data Science Journey gasar, inda ya dauki matsayi na 5 kuma ya sami 200 dubu rubles. Na farko sun biya miliyan daya, na biyu 500 dubu. Kudaden kyauta sun bambanta, kuma yanzu suna karuwa. Kasancewa a saman, zaku iya samun dubu 100 zuwa 500. Ina ajiye kuɗin kyauta don ilimi, wannan ita ce gudunmawata ga gaba, kuɗin da nake kashewa a rayuwar yau da kullum, na sami kaina.
Menene mafi ban sha'awa - mutum ko ƙungiyar hackathons?
- Idan muna magana ne game da haɓaka samfuri, to dole ne ya zama ƙungiya; mutum ɗaya ba zai iya yin shi ba. Kawai zai gaji kuma yana buƙatar tallafi. Amma idan muna magana, alal misali, game da AI Academy hackathon, to, aikin da ke can yana iyakance, babu buƙatar ƙirƙirar samfur. Sha'awar da ke can ta bambanta - don cim ma wani mutum wanda shi ma yana tasowa a wannan yanki.
- Ta yaya kuke shirin ci gaba? Ya kuke ganin sana'ar ku?
- Yanzu babban makasudin shine shirya aikin kimiyya mai mahimmanci, bincike, ta yadda zai bayyana a manyan tarurrukan kamar NeurIPS ko ICML - ML taron da ke gudana a ƙasashe daban-daban na duniya. Tambayar aiki a buɗe take, duba yadda ML ya haɓaka cikin shekaru 5 da suka gabata. Yana canzawa cikin sauri, yanzu yana da wuya a iya hasashen abin da zai biyo baya. Kuma idan muka yi magana game da ra'ayoyi da tsare-tsare ban da aikin kimiyya, to watakila zan ga kaina a cikin wani nau'i na aikin kaina, farawa a fagen AI da ML, amma wannan bai tabbata ba.
- A ra'ayin ku, menene iyakokin fasahar AI?
- To, gaba ɗaya, idan muka yi magana game da AI a matsayin wani abu da ke da wani nau'i na hankali, sarrafa bayanai, to, a nan gaba kadan, zai zama wani nau'i na fahimtar duniya da ke kewaye da mu. Idan muka yi magana game da hanyoyin sadarwa na jijiyoyi a cikin ilimin harshe na lissafi, alal misali, muna ƙoƙarin yin samfurin wani abu a gida, misali, harshe, ba tare da ba da samfurin fahimtar mahallin game da duniyarmu ba. Wato, idan mun sami damar shigar da wannan a cikin AI, za mu iya ƙirƙirar samfuran tattaunawa, bots ɗin taɗi waɗanda ba za su san ƙirar harshe kawai ba, amma kuma za su sami hangen nesa da sanin gaskiyar kimiyya. Kuma wannan shine abin da nake so in gani nan gaba.
Af, Cibiyar Ilimi ta Artificial Intelligence a halin yanzu tana daukar 'yan makaranta don sabon hackathon. Kuɗin kyauta kuma yana da mahimmanci, kuma aikin wannan shekara ya fi ban sha'awa - kuna buƙatar gina algorithm wanda ke hasashen kwarewar ɗan wasa bisa kididdigar wasan Dota 2 guda ɗaya. Don cikakkun bayanai, je zuwa .
source: www.habr.com