Klasterlash vazifasi - bu ma’lumotlarni guruhlarga taqsimlash: uyali aloqa
operatorining barcha mijozlarini to‘lov qobiliyati darajasi bo‘yicha
ajratish,
kosmik obyektlarni u yoki bu toifaga (sayyora, yulduz, qora tuynuk va h-k)
taqsimlash.
O‘lchovni qisqartirish vazifasi, keyinchalik ularni vizualizatsiya
qilish uchun
qulay bo‘lishi uchun (masalan, ma’lumotlarni siqish) ko‘p sonli xususiyatlarni
kichikroq (odatda 2-3) gacha kamaytirishdir.
Mustahkamlangan o‘qitish – bu ma’lum bir
vaziyatda rejani maksimal
darajada oshirish uchun tegishli choralarni ko‘rishdir. U ma’lum bir vaziyatda eng
yaxshi xatti-harakat yoki yo‘lni topish uchun turli xil dasturiy ta’minot va
mashinalar tomonidan qo‘llaniladi[2]. Mustahkamlangan o‘qitishni o‘qituvchisiz
o‘qitishdan shunday farq qiladiki, nazorat ostida o‘qitishda o‘quv ma’lumotlarida
javob kaliti bo‘ladi, shuning uchun modelning o‘zi to‘g‘ri javob bilan o‘qitiladi,
mustahkamlangan o‘qitishda esa javob yo‘q, lekin mustahkamlash bo‘limi nima
qilish kerakligini hal qiladi.
Bu o‘qitish usullari asosan Mashinali o‘qitish fanining asosiy qismi hisoblanadi va
bu o‘qitishlar orqali Mashinali o‘qitish tizimi shakllanadi.
International scientific conference "INFORMATION TECHNOLOGIES, NETWORKS AND
TELECOMMUNICATIONS" ITN&T-2022 Urgench, 2022y April 29-30
4
4
Foydalanilgan adabiyotlar.
1. Zhao, P. & Yu, B. (2006), ‘On model selection consistency of Lasso‘, Journal of
Machine Learning Research 7, 2541–2567. -235 p.
2. “Understanding Machine Learning: From Theory to Algorithms”, Shai Shalev-
Shwartz (The Hebrew University), Jerusalem Shai Ben-David (University of
Waterloo, Canada), 2014. -325 p.
3. Bessmertny I.A. Artificial intelligence - St. Petersburg: St. Petersburg State
University ITMO, 2010. - 132 p.
4. Brink H., Richards D., Feverolf M. Machine learning. - St. Petersburg: Peter, 2017. -
336 p.: ill. – (Series "Programmer’s Library").
5. China spent $279 billion on research and development last year USD – Access mode:
https://hightech.fm/2018/02/27/279-billion-on-rd
.