๋จธ์‹  ๋Ÿฌ๋‹
(Machine Learning)
Intro
์˜ˆ์‹œ : ์ฃผ์–ด์ง„ ์‚ฌ์ง„์œผ๋กœ๋ถ€ํ„ฐ ๊ณ ์–‘์ด ์‚ฌ์ง„์ธ์ง€ ๊ฐ•์•„์ง€ ์‚ฌ์ง„์ธ์ง€ ํŒ๋ณ„ํ•˜๋Š” ์ผ.
def prediction(์ด๋ฏธ์ง€ as input):
if ๋ˆˆ์ฝ”๊ท€๊ฐ€ ์žˆ์„ ๋•Œ :
if ๊ทผ๋ฐ ๊ฐ•์•„์ง€๋Š” ์•„๋‹ ๋•Œ
if ํ„ธ์ด ์žˆ๊ณ  ๊ผฌ๋ฆฌ๊ฐ€ ์žˆ์„ ๋•Œ :
if ๋‹ค๋ฅธ ๋™๋ฌผ์ด ์•„๋‹ ๋•Œ
โ€ฆ
์–ด์ผ€ํ•˜๋ˆ„โ€ฆ
return ๊ฒฐ๊ณผ
์ „ํ†ต์ ์ธ ๋ฐฉ์‹์œผ๋กœ ๊ฐœ๋ฐœ์ž๊ฐ€ ์ง์ ‘ ๊ทœ์น™์„ ์ •์˜
๋จธ์‹ ๋Ÿฌ๋‹(Machine Learning)?
โ€ข ๋ฐ์ดํ„ฐ๋ฅผ ํ†ตํ•ด ๊ธฐ๊ณ„๊ฐ€ ๊ทœ์น™์„ ์Šค์Šค๋กœ ์ฐพ์•„๋‚ด๋Š” ๋ฐฉ๋ฒ•
โ€ข ์ธ๊ณต์ง€๋Šฅ(AI)์˜ ํ•œ ๋ถ„์•ผ๋กœ, ๊ธฐ๊ณ„๊ฐ€ ํ•™์Šตํ•  ์ˆ˜ ์žˆ๋„๋ก ํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜๊ณผ ๊ธฐ์ˆ ์„ ๊ฐœ๋ฐœํ•˜๋Š” ๋ถ„์•ผ
์ธ๊ณต์ง€๋Šฅ(AI)?
์ธ๊ณต์ง€๋Šฅ(Artificial Intelligence)
โ€ข ๊ธฐ๊ณ„๊ฐ€ ์‚ฌ๋žŒ์ฒ˜๋Ÿผ ์ง€๋Šฅ์ด ํ•„์š”ํ•œ ์ผ์„ ํ•˜๋„๋ก ๋งŒ๋“œ๋Š” ๊ณผํ•™
โ€ข ํ™˜๊ฒฝ์„ ์ธ์ง€ํ•˜๊ณ  ๋ชฉํ‘œ๋ฅผ ์„ฑ๊ณต์ ์œผ๋กœ ๋‹ฌ์„ฑํ•  ๊ฐ€๋Šฅ์„ฑ์„ ๊ทน๋Œ€ํ™”ํ•˜๋„๋ก ์กฐ์น˜๋ฅผ ์ทจํ•  ์ˆ˜ ์žˆ๋Š” ์‹œ์Šคํ…œ
Ex) ์ž์œจ ์ฃผํ–‰ ์ž๋™์ฐจ, ์Šค๋งˆํŠธ ์Šคํ”ผ์ปค, ์ฑ—๋ด‡, ์ธ๊ณต์ง€๋Šฅ ๋กœ๋ด‡, ์ด๋ฏธ์ง€ ์ธ์‹, ๊ฐœ์ธํ™” ์ถ”์ฒœ, ๊ธฐ๊ณ„ ๋ฒˆ์—ญ
๋”ฅ๋Ÿฌ๋‹(DL)?
๋”ฅ ๋Ÿฌ๋‹(Deep Learning)
โ€ข ์ธ๊ฐ„์˜ ๋‡Œ์—์„œ ์‹ ๊ฒฝ ์„ธํฌ๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ๋ฐฉ์‹๊ณผ ์œ ์‚ฌํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์‚ฌ์šฉํ•˜๋Š” ๋จธ์‹  ๋Ÿฌ๋‹(ML) ์˜ ํ•˜์œ„ ๋ถ„์•ผ
โ€ข ์šฐ๋ฆฌ ์ƒํ™œ์˜ ์˜ˆ๋ฅผ ํ†ตํ•ด ์šฐ๋ฆฌ ์ธ๊ฐ„๋“ค์ด ์‰ฝ๊ณ  ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ํ•˜๋Š” ์ผ์„ ์ปดํ“จํ„ฐ์— ๊ฐ€๋ฅด์น˜๋Š” ๋จธ์‹  ๋Ÿฌ๋‹ ๊ธฐ์ˆ 
์ธ๊ณต ์‹ ๊ฒฝ๋ง(Artificial Neural Network)
โ€ข ์ธ๊ฐ„์˜ ๋‡Œ ์† ๋‰ด๋Ÿฐ์˜ ์ž‘์šฉ์„ ๋ณธ๋–  ํŒจํ„ด์„
๊ตฌ์„ฑํ•œ ์ปดํ“จํŒ… ์‹œ์Šคํ…œ์˜ ์ผ์ข…
๋”ฅ๋Ÿฌ๋‹(DL)?
์‹ฌ์ธต ์‹ ๊ฒฝ๋ง(DNN, Deep Neural Network)
โ€ข ๋ ˆ์ด์–ด๊ฐ€ ํ•œ ์ธต์œผ๋กœ๋งŒ ๊ตฌ์„ฑ๋œ ๊ฒƒ์ด ์•„๋‹ˆ๋ผ
์—ฌ๋Ÿฌ ์ธต, ๋‹ค์‹œ ๋งํ•ด ๊นŠ์€ ์ธต์œผ๋กœ ๊ตฌ์„ฑ๋œ ์ธ๊ณต ์‹ ๊ฒฝ๋ง
์‹ฌ์ธต ์‹ ๊ฒฝ๋ง์ด ํ•™์Šตํ•˜๋Š” ๊ณผ์ • -> ๋”ฅ๋Ÿฌ๋‹
ใ€Œ ใ€
์ธ๊ณต์ง€๋Šฅ(AI)? ๋จธ์‹ ๋Ÿฌ๋‹(ML)? ๋”ฅ๋Ÿฌ๋‹(DL)?
์ธ๊ณต์ง€๋Šฅ(AI) > ๋จธ์‹ ๋Ÿฌ๋‹(ML) > ๋”ฅ๋Ÿฌ๋‹(DL)
ํ”„๋กœ๊ทธ๋ž˜๋ฐ ๋ฐฉ์‹ & ๋จธ์‹ ๋Ÿฌ๋‹ ๋ฐฉ์‹
๊ธฐ์กด ์ปดํ“จํ„ฐ ํ”„๋กœ๊ทธ๋žจ ๋จธ์‹  ๋Ÿฌ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜
์ž…๋ ฅ๊ฐ’์„ ๋ฐ›์œผ๋ฉด ๊ฒฐ๊ณผ๊ฐ’์„ ์ถœ๋ ฅํ•˜๋Š” ๋ฐฉ์‹
๋‚ด๋ถ€์  ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด ์ค‘์š”
์ž…๋ ฅ๊ฐ’๊ณผ ๊ฒฐ๊ณผ๊ฐ’์„ ๊ฐ€์ง€๊ณ 
์ปดํ“จํ„ฐ๊ฐ€ ํ•™์Šต์„ ํ†ตํ•˜์—ฌ ๊ทœ์น™์„ ๋„์ถœ ํ•ด๋‚ธ ๋‹ค์Œ
์ƒˆ๋กœ์šด ์ž…๋ ฅ๊ฐ’์ด ์ฃผ์–ด์กŒ์„ ๋•Œ ๊ฒฐ๊ณผ๋ฅผ ์˜ˆ์ธก
๋ฐ์ดํ„ฐ๊ฐ€ ์ค‘์š”. ๋ฐ์ดํ„ฐ ์ค‘์‹ฌ & ๋ชจ๋ธ ์ค‘์‹ฌ
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ์ง€๋„ํ•™์Šต
๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์‹œํ‚ฌ ๋•Œ ๋‹ต๊นŒ์ง€ ๊ฐ™์ด ์•Œ๋ ค์ฃผ๋Š” ๋ฐฉ์‹์˜ ํ•™์Šต ๋ฐฉ๋ฒ•
๊ฐ•์•„์ง€์™€ ๊ณ ์–‘์ด ์‚ฌ์ง„์„ ๊ตฌ๋ถ„ํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ํ•™์Šตํ•  ๋•Œ
์‚ฌ์ง„๊ณผ ํ•จ๊ป˜ ์ด๊ฒƒ์ด ๊ฐ•์ด์ง€์ธ์ง€ ๊ณ ์–‘์ด์ธ์ง€ ์•Œ๋ ค์ฃผ๋Š” ๊ฒƒ
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ์ง€๋„ํ•™์Šต
โ€ข ๋ ˆ์ด๋ธ”์ด ๋‹ฌ๋ฆฐ ๋ฐ์ดํ„ฐ๋กœ ํ•™์Šต ํ›„ ์ƒˆ๋กœ์šด ๋ฐ์ดํ„ฐ๊ฐ€
์–ด๋А ๊ทธ๋ฃน์— ์†ํ•˜๋Š”์ง€ ์ฐพ์•„๋‚ด๋Š” ๋ฐฉ๋ฒ•
โ€ข ์˜ˆ์ธกํ•˜๋ ค๋Š” ๊ฐ’์ด ์ด์‚ฐ ๊ฐ’ ์ผ ๊ฒฝ์šฐ (์ž๋™์ฐจ, ์‚ฌ๋žŒ, ๋„๋กœ)
โ€ข ์˜ˆ๋กœ โ€˜์ŠคํŒธ ๋ฉ”์ผ ๊ตฌ๋ถ„โ€™
โ€ข ์—ฐ์†์ ์ธ ์ˆซ์ž(์‹ค์ˆ˜)๋ฅผ ์˜ˆ์ธกํ•˜๋Š” ๊ฒƒ
โ€ข ์˜ˆ์ธกํ•˜๋ ค๋Š” ๊ฐ’์ด ์—ฐ์† ๊ฐ’ ์ผ ๊ฒฝ์šฐ (0.31, 0.301, 0.3001)
โ€ข ์˜ˆ๋กœ โ€˜์•„ํŒŒํŠธ๊ฐ€๊ฒฉ ์˜ˆ์ธกโ€™, โ€˜๊ธฐ์˜จ ์˜ˆ์ธกโ€™
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต
๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์‹œํ‚ฌ ๋•Œ ๋‹ต์„ ์•Œ๋ ค์ฃผ์ง€ ์•Š๊ณ , ๋ฐ์ดํ„ฐ์˜ ํŠน์„ฑ์— ๋”ฐ๋ผ
๊ด€๊ณ„๋‚˜ ํŒจํ„ด์„ ์ฐพ์•„ ์Šค์Šค๋กœ ํ•™์Šตํ•˜๋Š” ๋ฐฉ์‹
๋™๋ฌผ ์‚ฌ์ง„์„ ๋ถ„๋ฅ˜ํ•  ๊ฒฝ์šฐ ์‚ฌ์ง„์˜ ๊ตฌ์กฐ๋‚˜ ํŠน์„ฑ ๋ณ„๋กœ ๋ถ„๋ฅ˜
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต
๊ตฐ์ง‘ํ™” (Clustering)
โ€ข ๋ ˆ์ด๋ธ”์ด ์—†๋Š” ํ•™์Šต ๋ฐ์ดํ„ฐ๋“ค์˜ ํŠน์ง•์„ ๋ถ„์„ํ•˜์—ฌ ์œ ์‚ฌํ•œ ํŠน์ง•์„ ๊ฐ€์ง„ ๋ฐ์ดํ„ฐ ๋ผ๋ฆฌ
๊ทธ๋ฃนํ™” ํ•˜๋Š” ๊ฒƒ
๊ตฐ์ง‘ํ™”์˜ ๊ณผ์ •
โ€ข ๋ฐ์ดํ„ฐ๋ฅผ ์ขŒํ‘œํ™” -> ๊ฐ€๊นŒ์šด ๊ฒƒ๋“ค ๋ผ๋ฆฌ ๋ชจ์•„ ํ•˜๋‚˜์˜ ๊ทธ๋ฃน์„ ๋งŒ๋“ ๋‹ค
( ์ขŒํ‘œ์ƒ ๊ฐ€๊น๋‹ค๋Š” ๊ฒƒ์€ ๋ฐ์ดํ„ฐ๊ฐ€ ๋น„์Šทํ•˜๋‹ค๋Š” ๊ฒƒ์„ ์˜๋ฏธ)
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต
์—ฐ๊ด€ ๊ทœ์น™ (association rule)
โ€ข ์„œ๋กœ ์—ฐ๊ด€๋œ ํŠน์ง•์„ ์ฐพ์•„ ๋‚ด๋Š” ๊ฒƒ
โ€ข ์ฃผ๋กœ ์ถ”์ฒœ๊ณผ ๊ด€๋ จ๋œ ๋ถ€๋ถ„์— ์‚ฌ์šฉ (์žฅ๋ฐ”๊ตฌ๋‹ˆ ๋ถ„์„)
โ€ข ๋ผ๋ฉด์„ ๊ตฌ์ž…ํ•œ ์‚ฌ๋žŒ์€ ๊ณ„๋ž€์„ ๊ตฌ์ž…ํ•  ํ™•๋ฅ ์ด ๋†’๋‹ค
โ€ข ์ฆ‰, ๋ผ๋ฉด๊ณผ ๊ณ„๋ž€์€ ์„œ๋กœ ์—ฐ๊ด€์„ฑ(Association)์ด ๋†’๋‹ค
์—ฐ๊ด€์„ฑ์„ ํŒŒ์•… -> ์—ฐ๊ด€์„ฑ์ด ์žˆ๋Š” ์ƒํ’ˆ์„ ์ถ”์ฒœ
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต
๊ตฐ์ง‘ํ™” vs ์—ฐ๊ด€ ๊ทœ์น™
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๊ฐ•ํ™”ํ•™์Šต
ํ–‰๋™์— ๋Œ€ํ•œ ๋ณด์ƒ์„ ๋ฐ›์œผ๋ฉด์„œ ํ•™์Šตํ•˜์—ฌ ์–ด๋–ค ํ™˜๊ฒฝ ์•ˆ์—์„œ ์„ ํƒ ๊ฐ€๋Šฅํ•œ
ํ–‰๋™๋“ค ์ค‘ ๋ณด์ƒ์„ ์ตœ๋Œ€ํ™”ํ•˜๋Š” ํ–‰๋™ ๋˜๋Š” ํ–‰๋™ ์ˆœ์„œ๋ฅผ ์„ ํƒํ•˜๋Š” ๋ฐฉ๋ฒ•
์ฆ‰, ์–ด๋–ค ํ™˜๊ฒฝ ์•ˆ์—์„œ ์ •์˜๋œ ์ฃผ์ฒด(agent)๊ฐ€ ํ˜„์žฌ์˜ ์ƒํƒœ(state)๋ฅผ ๊ด€์ฐฐํ•˜์—ฌ
์„ ํƒํ•  ์ˆ˜ ์žˆ๋Š” ํ–‰๋™(action)๋“ค ์ค‘์—์„œ ๊ฐ€์žฅ ์ตœ๋Œ€์˜ ๋ณด์ƒ(reward)์„ ๊ฐ€์ ธ๋‹ค์ฃผ๋Š”
ํ–‰๋™์ด ๋ฌด์—‡์ธ์ง€๋ฅผ ํ•™์Šตํ•˜๋Š” ๊ฒƒ (ex ์•ŒํŒŒ๊ณ , ๊ฐ์ข… ๊ฒŒ์ž„)
โ€ข ํ™˜๊ฒฝ(environment)
โ€ข ์—์ด์ „ํŠธ(agent)
โ€ข ์ƒํƒœ(state)
โ€ข ํ–‰๋™(action)
โ€ข ๋ณด์ƒ(reward)
โ€ข ์ •์ฑ…(policy)
๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๊ฐ•ํ™”ํ•™์Šต
๊ฐ•ํ™”ํ•™์Šต์˜ ์˜ˆ
๋” ๋งŽ์€ ๋ณด์ƒ์„ ๋ฐ›์„ ์ˆ˜ ์žˆ๋Š” ์ •์ฑ…์„ ๋งŒ๋“œ๋Š” ๊ฒƒ์ด ํ•ต์‹ฌ
โ€ข ๊ฒŒ์ž„ -> ํ™˜๊ฒฝ(environment)
โ€ข ๊ฒŒ์ด๋จธ -> ์—์ด์ „ํŠธ(agent)
โ€ข ๊ฒŒ์ž„ํ™”๋ฉด -> ์ƒํƒœ(state)
โ€ข ๊ฒŒ์ด๋จธ์˜ ์กฐ์ž‘ -> ํ–‰๋™(action)
โ€ข ์ƒ๊ณผ ๋ฒŒ -> ๋ณด์ƒ(reward)
โ€ข ๊ฒŒ์ด๋จธ์˜ ํŒ๋‹จ๋ ฅ -> ์ •์ฑ…(policy)
1. ๊ฒŒ์ž„์€ ๊ฒŒ์ด๋จธ์—๊ฒŒ ๊ฒŒ์ž„ํ™”๋ฉด์„ ํ†ตํ•ด ํ˜„์žฌ์˜ ์ƒํƒœ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค.
(์บ๋ฆญํ„ฐ์˜ ์œ„์น˜๋‚˜ ์žฅ์• ๋ฌผ์˜ ์œ„์น˜ ๋“ฑ๋“ฑ)
2. ํ˜„์žฌ์˜ ์ ์ˆ˜๋„ ์•Œ๋ ค์ค€ ๋’ค ๊ฒŒ์ด๋จธ์—๊ฒŒ๋Š” ์ ์ˆ˜๊ฐ€ ๋†’์•„์ง€๋Š” ๊ฒƒ์ด
์ƒ์ด๊ณ  ๋ฐ˜๋Œ€๋กœ ์ ์ˆ˜๊ฐ€ ๋‚ฎ์•„์ง€๋Š” ๊ฒƒ์ด ๋ฒŒ์ด๋‹ค.
3. ํ”Œ๋ ˆ์ด(๊ด€์ฐฐ)๋ฅผ ํ•˜๋ฉฐ ์–ด๋–ป๊ฒŒ ํ•˜๋ฉด ๋” ๋งŽ์€ ๋ณด์ƒ์„ ์–ป๊ฒŒ ๋˜๋Š”์ง€,
๋˜ํ•œ ๋” ์ ์€ ๋ฒŒ์„ ๋ฐ›๊ฒŒ ๋˜๋Š”์ง€ ์•Œ๊ฒŒ ๋œ๋‹ค.
4. ๊ทธ๋กœ ์ธํ•ด ๊ฒŒ์ด๋จธ์˜ ํŒ๋‹จ๋ ฅ์ด ๊ฐ•ํ™”๋œ๋‹ค.
5. ๊ฐ•ํ™”๋œ ํŒ๋‹จ๋ ฅ์— ๋”ฐ๋ผ ๊ฒŒ์ž„์„ ์กฐ์ž‘ํ•œ๋‹ค
6. ๊ทธ ์กฐ์ž‘์€ ๊ฒŒ์ž„์— ๋ณ€ํ™”๋ฅผ ๊ฐ€์ ธ์˜จ๋‹ค.
๋จธ์‹ ๋Ÿฌ๋‹์ด ์“ฐ์ด๋Š” ๋ถ„์•ผ
๋จธ์‹ ๋Ÿฌ๋‹ ์„œ๋น„์Šค
Amazon SageMaker
MNIST Dataset
MNIST Dataset & Fashion MNIST Dataset
โ€ข 28ร—28์˜ ํฌ๊ธฐ์˜ 0~9 ์ˆซ์ž ์ด๋ฏธ์ง€ 70,000๊ฐœ
MNIST Dataset & Fashion MNIST Dataset
โ€ข 28ร—28์˜ ํฌ๊ธฐ์˜ ํ”ฝ์…€ ์ด๋ฏธ์ง€ 70,000๊ฐœ
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot
๋ฐ์ดํ„ฐ ์˜ˆ์ธก
๋ฐ์ดํ„ฐ ์ค€๋น„ โ†’ ํ•™์Šต โ†’ ๋ฐฐํฌ โ†’ ํ…Œ์ŠคํŠธ โ†’ ์˜ˆ์ธก
End
Thank You
์ž๋ฃŒ ์ถœ์ฒ˜
https://blog.lgcns.com/2212
https://hongong.hanbit.co.kr/ai-%EB%AC%B4%EC%97%87%EC%9D%B8%EA%B0%80-%EC%9D%B8%EA%B3%B5%EC%A7%80%EB%8A%A5-
%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D-%EB%94%A5%EB%9F%AC%EB%8B%9D-%EC%B0%A8%EC%9D%B4%EC%A0%90-
%EC%B4%9D%EC%A0%95%EB%A6%AC/
https://wikidocs.net/21679
https://codong.tistory.com/37
https://blog.illunex.com/%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D%EA%B8%B0%EA%B3%84%ED%95%99%EC%8A%B5%EC%97%90-
%EA%B4%80%ED%95%98%EC%97%AC-1%ED%83%84/
https://opentutorials.org/module/4916/28934
https://school.coding-x.com/lesson/138
https://kimeunh3.github.io/machine%20learning/ml_05/
https://labs.brandi.co.kr/2018/05/17/ohyj.html

More Related Content

PDF
[ํ™๋Œ€ ๋จธ์‹ ๋Ÿฌ๋‹ ์Šคํ„ฐ๋”” - ํ•ธ์ฆˆ์˜จ ๋จธ์‹ ๋Ÿฌ๋‹] 1์žฅ. ํ•œ๋ˆˆ์— ๋ณด๋Š” ๋จธ์‹ ๋Ÿฌ๋‹
PDF
ํ•œ๋ˆˆ์— ๋ณด๋Š” ๋จธ์‹ ๋Ÿฌ๋‹
PPTX
๋ฐ์ดํ„ฐ๊ณผํ•™ ์ž…๋ฌธ 8์žฅ
PDF
์•ˆ.์ „.์ œ.์ผ. ๊ฐ•ํ™”ํ•™์Šต!
PPTX
๋งํฌ๋“œ์ธ์˜ Big Data Recommendation Products - ์–ด์ œ์˜ ๋ฐ์ดํ„ฐ๋ฅผ ํ†ตํ•ด ๋‚ด์ผ์„ ์˜ˆ์ธกํ•œ๋‹ค
PDF
๋จธ์‹ ๋Ÿฌ๋‹ ๋ฐ ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ์ดˆ (2020.01.07)
PDF
AI_introduction and requirements(2024.05.12).pdf
PDF
T ka kr_4th
[ํ™๋Œ€ ๋จธ์‹ ๋Ÿฌ๋‹ ์Šคํ„ฐ๋”” - ํ•ธ์ฆˆ์˜จ ๋จธ์‹ ๋Ÿฌ๋‹] 1์žฅ. ํ•œ๋ˆˆ์— ๋ณด๋Š” ๋จธ์‹ ๋Ÿฌ๋‹
ํ•œ๋ˆˆ์— ๋ณด๋Š” ๋จธ์‹ ๋Ÿฌ๋‹
๋ฐ์ดํ„ฐ๊ณผํ•™ ์ž…๋ฌธ 8์žฅ
์•ˆ.์ „.์ œ.์ผ. ๊ฐ•ํ™”ํ•™์Šต!
๋งํฌ๋“œ์ธ์˜ Big Data Recommendation Products - ์–ด์ œ์˜ ๋ฐ์ดํ„ฐ๋ฅผ ํ†ตํ•ด ๋‚ด์ผ์„ ์˜ˆ์ธกํ•œ๋‹ค
๋จธ์‹ ๋Ÿฌ๋‹ ๋ฐ ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ์ดˆ (2020.01.07)
AI_introduction and requirements(2024.05.12).pdf
T ka kr_4th

Similar to MachineLearning (20)

PDF
๋จธ์‹ ๋Ÿฌ๋‹(๋”ฅ๋Ÿฌ๋‹ ์š”์•ฝ)
PPTX
แ„‹แ…ตแ†ซแ„€แ…ฉแ†ผแ„Œแ…ตแ„‚แ…ณแ†ผ แ„†แ…กแ†บแ„‡แ…ฉ๊ธฐ
PDF
์บ๋นˆ๋จธํ”ผ ๋จธ์‹ ๋Ÿฌ๋‹ Kevin Murphy Machine Learning Statistic
PDF
์•ŒํŒŒ๊ณ  ํ’€์–ด๋ณด๊ธฐ / Alpha Technical Review
PDF
PDF
ํŒŒ์ด์ฌ์œผ๋กœ ์ตํžˆ๋Š” ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ๋ณธ (18๋…„)
PDF
00_ML ๊ธฐ์ดˆ_๊ธฐ๋ณธ๊ฐœ๋…
PDF
์ฐฝ๋ฐœ ์„ธ๋ฏธ๋‚˜ - Bing์—์„œ์˜ ๋จธ์‹ ๋Ÿฌ๋‹
PPTX
1.introduction(epoch#2)
PDF
NDC 2016 ๊น€์ •์ฃผ - ๊ธฐ๊ณ„ํ•™์Šต์„ ํ™œ์šฉํ•œ ๊ฒŒ์ž„์–ด๋ทฐ์ง• ๊ฒ€์ถœ
PDF
2016 kcd ์„ธ๋ฏธ๋‚˜ ๋ฐœํ‘œ์ž๋ฃŒ. ๊ตฌ๊ธ€ํฌํ† ๋กœ ๋ฐ”๋ผ๋ณธ ์ธ๊ณต์ง€๋Šฅ๊ณผ ๋จธ์‹ ๋Ÿฌ๋‹
PPT
NNKproject Korean version
PDF
Mu zero review by nomoreid
PDF
Machine learning bysogood
PDF
[D2 CAMPUS] Tech meet-up `data science` ๋ฐœํ‘œ์ž๋ฃŒ
PDF
what is deep learning?
PDF
แ„€แ…ตแ„€แ…จแ„’แ…กแ†จแ„‰แ…ณแ†ธแ„‹แ…ณแ†ฏ แ„‹แ…ตแ„‹แ…ญแ†ผแ„’แ…กแ†ซ แ„‰แ…ฎแ†บแ„Œแ…กแ„‹แ…ตแ†ซแ„‰แ…ตแ†จแ„€แ…ต แ„Œแ…ฆแ„Œแ…กแ†จ
PPTX
Chapter 11 Practical Methodology
PDF
ํ…์„œํ”Œ๋กœ์šฐ๋กœ ๋ฐฐ์šฐ๋Š” ๋”ฅ๋Ÿฌ๋‹
PDF
CS294-112 Lecture 13
๋จธ์‹ ๋Ÿฌ๋‹(๋”ฅ๋Ÿฌ๋‹ ์š”์•ฝ)
แ„‹แ…ตแ†ซแ„€แ…ฉแ†ผแ„Œแ…ตแ„‚แ…ณแ†ผ แ„†แ…กแ†บแ„‡แ…ฉ๊ธฐ
์บ๋นˆ๋จธํ”ผ ๋จธ์‹ ๋Ÿฌ๋‹ Kevin Murphy Machine Learning Statistic
์•ŒํŒŒ๊ณ  ํ’€์–ด๋ณด๊ธฐ / Alpha Technical Review
ํŒŒ์ด์ฌ์œผ๋กœ ์ตํžˆ๋Š” ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ๋ณธ (18๋…„)
00_ML ๊ธฐ์ดˆ_๊ธฐ๋ณธ๊ฐœ๋…
์ฐฝ๋ฐœ ์„ธ๋ฏธ๋‚˜ - Bing์—์„œ์˜ ๋จธ์‹ ๋Ÿฌ๋‹
1.introduction(epoch#2)
NDC 2016 ๊น€์ •์ฃผ - ๊ธฐ๊ณ„ํ•™์Šต์„ ํ™œ์šฉํ•œ ๊ฒŒ์ž„์–ด๋ทฐ์ง• ๊ฒ€์ถœ
2016 kcd ์„ธ๋ฏธ๋‚˜ ๋ฐœํ‘œ์ž๋ฃŒ. ๊ตฌ๊ธ€ํฌํ† ๋กœ ๋ฐ”๋ผ๋ณธ ์ธ๊ณต์ง€๋Šฅ๊ณผ ๋จธ์‹ ๋Ÿฌ๋‹
NNKproject Korean version
Mu zero review by nomoreid
Machine learning bysogood
[D2 CAMPUS] Tech meet-up `data science` ๋ฐœํ‘œ์ž๋ฃŒ
what is deep learning?
แ„€แ…ตแ„€แ…จแ„’แ…กแ†จแ„‰แ…ณแ†ธแ„‹แ…ณแ†ฏ แ„‹แ…ตแ„‹แ…ญแ†ผแ„’แ…กแ†ซ แ„‰แ…ฎแ†บแ„Œแ…กแ„‹แ…ตแ†ซแ„‰แ…ตแ†จแ„€แ…ต แ„Œแ…ฆแ„Œแ…กแ†จ
Chapter 11 Practical Methodology
ํ…์„œํ”Œ๋กœ์šฐ๋กœ ๋ฐฐ์šฐ๋Š” ๋”ฅ๋Ÿฌ๋‹
CS294-112 Lecture 13
Ad

More from Wonjun Hwang (20)

PDF
20250822_Kit-Works Tram Study_Spring Actuator.pdf
PDF
20250822_Kit-Works Team Study_ํ•จ๊ป˜์ž๋ผ๊ธฐ-1.pdf
PDF
20250802 _ TOSS MAKERS CONFERENCE 25.pdf
PDF
20250725_Kit-Works Team Study_GOOGLE I_O 2025.pdf
PPTX
20250725_Kit-Works Team Study_Spring AI.pptx
PPTX
20250718_Next.js๋ฅผ ๋– ๋‚˜๋Š” ๊ฐœ๋ฐœ์ž๋“ค: ๋น„ํŒ๊ณผ ๋Œ€์•ˆ ํ”„๋ ˆ์ž„์›Œํฌ ๋ถ„์„.pptx
PDF
Kit-Works Team Study_20250718_์ž๋ฐ”์˜-enum.pdf
PDF
Kit-Works Team Study_Vibe Coding ๋„์ „ํ•ด๋ณด๊ธฐ.pdf
PPTX
Kit-Works Team Study_๋ธŒ๋ผ์šฐ์ € ๊ฒ€์ƒ‰ ๊ณผ์ •_20250704_์†๋ฌธ์ˆ˜.pptx
PDF
Kit-Works Team Study_20250627_ํ•œ๋‹ฌ๋งŒ์—๋งŒ๋“ ์‚ฌ๋‚ด์„œ๋น„์Šคํ‚ค๋ง(์–‘๋‹ค์œ—).pdf
PDF
Kit-Works Team Study_20250627_๊ธฐ์ˆ  ๋ถ€์ฑ„_๊น€๊ฒฝ์ˆ˜.pdf
PPTX
20250530_Kit-Works Team Study_๊ฒฐ์ œ, ๋„ˆ ๋ฏฟ์–ด๋„ ๋ ๊นŒ.pptx
PPTX
20250620_Kit-Works Team Study_jspecify.pptx
PDF
20250523_Kit-Works Team Study_์œค์ •๋นˆ_๋†“์น˜๊ณ  ์žˆ๋˜ ์›น ์ ‘๊ทผ์„ฑ.pdf
PPTX
20250523_Kit-Works Team Study_Exception.pptx
PPTX
Kit-Works Team Study-20250517_uuid_๊น€ํ•œ๋‚˜.pptx
PDF
Kit-Works Team Study_20240517_์žฅํ˜„์ •_Claude์—์„œMCP์‚ฌ์šฉํ•ด๋ณด๊ธฐ.pdf
PDF
Kit-Works Team Study_์•„์ง๋„ Dockefile.pdf_๊น€์„ฑํ˜ธ
PDF
Kit-Works Team Study_ํŒ€์Šคํ„ฐ๋””_๊น€ํ•œ์†”_nuqs_20250509.pdf
PDF
Kit-Works Team Study_๊ณตํ—ˆ์ฐธ,๋ถ€์กด์žฌ์ฆ๋ช…,ํŠธ๋Ÿฌ๋ธ”์ŠˆํŒ….pdf
20250822_Kit-Works Tram Study_Spring Actuator.pdf
20250822_Kit-Works Team Study_ํ•จ๊ป˜์ž๋ผ๊ธฐ-1.pdf
20250802 _ TOSS MAKERS CONFERENCE 25.pdf
20250725_Kit-Works Team Study_GOOGLE I_O 2025.pdf
20250725_Kit-Works Team Study_Spring AI.pptx
20250718_Next.js๋ฅผ ๋– ๋‚˜๋Š” ๊ฐœ๋ฐœ์ž๋“ค: ๋น„ํŒ๊ณผ ๋Œ€์•ˆ ํ”„๋ ˆ์ž„์›Œํฌ ๋ถ„์„.pptx
Kit-Works Team Study_20250718_์ž๋ฐ”์˜-enum.pdf
Kit-Works Team Study_Vibe Coding ๋„์ „ํ•ด๋ณด๊ธฐ.pdf
Kit-Works Team Study_๋ธŒ๋ผ์šฐ์ € ๊ฒ€์ƒ‰ ๊ณผ์ •_20250704_์†๋ฌธ์ˆ˜.pptx
Kit-Works Team Study_20250627_ํ•œ๋‹ฌ๋งŒ์—๋งŒ๋“ ์‚ฌ๋‚ด์„œ๋น„์Šคํ‚ค๋ง(์–‘๋‹ค์œ—).pdf
Kit-Works Team Study_20250627_๊ธฐ์ˆ  ๋ถ€์ฑ„_๊น€๊ฒฝ์ˆ˜.pdf
20250530_Kit-Works Team Study_๊ฒฐ์ œ, ๋„ˆ ๋ฏฟ์–ด๋„ ๋ ๊นŒ.pptx
20250620_Kit-Works Team Study_jspecify.pptx
20250523_Kit-Works Team Study_์œค์ •๋นˆ_๋†“์น˜๊ณ  ์žˆ๋˜ ์›น ์ ‘๊ทผ์„ฑ.pdf
20250523_Kit-Works Team Study_Exception.pptx
Kit-Works Team Study-20250517_uuid_๊น€ํ•œ๋‚˜.pptx
Kit-Works Team Study_20240517_์žฅํ˜„์ •_Claude์—์„œMCP์‚ฌ์šฉํ•ด๋ณด๊ธฐ.pdf
Kit-Works Team Study_์•„์ง๋„ Dockefile.pdf_๊น€์„ฑํ˜ธ
Kit-Works Team Study_ํŒ€์Šคํ„ฐ๋””_๊น€ํ•œ์†”_nuqs_20250509.pdf
Kit-Works Team Study_๊ณตํ—ˆ์ฐธ,๋ถ€์กด์žฌ์ฆ๋ช…,ํŠธ๋Ÿฌ๋ธ”์ŠˆํŒ….pdf
Ad

MachineLearning

  • 2. Intro ์˜ˆ์‹œ : ์ฃผ์–ด์ง„ ์‚ฌ์ง„์œผ๋กœ๋ถ€ํ„ฐ ๊ณ ์–‘์ด ์‚ฌ์ง„์ธ์ง€ ๊ฐ•์•„์ง€ ์‚ฌ์ง„์ธ์ง€ ํŒ๋ณ„ํ•˜๋Š” ์ผ. def prediction(์ด๋ฏธ์ง€ as input): if ๋ˆˆ์ฝ”๊ท€๊ฐ€ ์žˆ์„ ๋•Œ : if ๊ทผ๋ฐ ๊ฐ•์•„์ง€๋Š” ์•„๋‹ ๋•Œ if ํ„ธ์ด ์žˆ๊ณ  ๊ผฌ๋ฆฌ๊ฐ€ ์žˆ์„ ๋•Œ : if ๋‹ค๋ฅธ ๋™๋ฌผ์ด ์•„๋‹ ๋•Œ โ€ฆ ์–ด์ผ€ํ•˜๋ˆ„โ€ฆ return ๊ฒฐ๊ณผ ์ „ํ†ต์ ์ธ ๋ฐฉ์‹์œผ๋กœ ๊ฐœ๋ฐœ์ž๊ฐ€ ์ง์ ‘ ๊ทœ์น™์„ ์ •์˜
  • 3. ๋จธ์‹ ๋Ÿฌ๋‹(Machine Learning)? โ€ข ๋ฐ์ดํ„ฐ๋ฅผ ํ†ตํ•ด ๊ธฐ๊ณ„๊ฐ€ ๊ทœ์น™์„ ์Šค์Šค๋กœ ์ฐพ์•„๋‚ด๋Š” ๋ฐฉ๋ฒ• โ€ข ์ธ๊ณต์ง€๋Šฅ(AI)์˜ ํ•œ ๋ถ„์•ผ๋กœ, ๊ธฐ๊ณ„๊ฐ€ ํ•™์Šตํ•  ์ˆ˜ ์žˆ๋„๋ก ํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜๊ณผ ๊ธฐ์ˆ ์„ ๊ฐœ๋ฐœํ•˜๋Š” ๋ถ„์•ผ
  • 4. ์ธ๊ณต์ง€๋Šฅ(AI)? ์ธ๊ณต์ง€๋Šฅ(Artificial Intelligence) โ€ข ๊ธฐ๊ณ„๊ฐ€ ์‚ฌ๋žŒ์ฒ˜๋Ÿผ ์ง€๋Šฅ์ด ํ•„์š”ํ•œ ์ผ์„ ํ•˜๋„๋ก ๋งŒ๋“œ๋Š” ๊ณผํ•™ โ€ข ํ™˜๊ฒฝ์„ ์ธ์ง€ํ•˜๊ณ  ๋ชฉํ‘œ๋ฅผ ์„ฑ๊ณต์ ์œผ๋กœ ๋‹ฌ์„ฑํ•  ๊ฐ€๋Šฅ์„ฑ์„ ๊ทน๋Œ€ํ™”ํ•˜๋„๋ก ์กฐ์น˜๋ฅผ ์ทจํ•  ์ˆ˜ ์žˆ๋Š” ์‹œ์Šคํ…œ Ex) ์ž์œจ ์ฃผํ–‰ ์ž๋™์ฐจ, ์Šค๋งˆํŠธ ์Šคํ”ผ์ปค, ์ฑ—๋ด‡, ์ธ๊ณต์ง€๋Šฅ ๋กœ๋ด‡, ์ด๋ฏธ์ง€ ์ธ์‹, ๊ฐœ์ธํ™” ์ถ”์ฒœ, ๊ธฐ๊ณ„ ๋ฒˆ์—ญ
  • 5. ๋”ฅ๋Ÿฌ๋‹(DL)? ๋”ฅ ๋Ÿฌ๋‹(Deep Learning) โ€ข ์ธ๊ฐ„์˜ ๋‡Œ์—์„œ ์‹ ๊ฒฝ ์„ธํฌ๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ๋ฐฉ์‹๊ณผ ์œ ์‚ฌํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์‚ฌ์šฉํ•˜๋Š” ๋จธ์‹  ๋Ÿฌ๋‹(ML) ์˜ ํ•˜์œ„ ๋ถ„์•ผ โ€ข ์šฐ๋ฆฌ ์ƒํ™œ์˜ ์˜ˆ๋ฅผ ํ†ตํ•ด ์šฐ๋ฆฌ ์ธ๊ฐ„๋“ค์ด ์‰ฝ๊ณ  ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ํ•˜๋Š” ์ผ์„ ์ปดํ“จํ„ฐ์— ๊ฐ€๋ฅด์น˜๋Š” ๋จธ์‹  ๋Ÿฌ๋‹ ๊ธฐ์ˆ  ์ธ๊ณต ์‹ ๊ฒฝ๋ง(Artificial Neural Network) โ€ข ์ธ๊ฐ„์˜ ๋‡Œ ์† ๋‰ด๋Ÿฐ์˜ ์ž‘์šฉ์„ ๋ณธ๋–  ํŒจํ„ด์„ ๊ตฌ์„ฑํ•œ ์ปดํ“จํŒ… ์‹œ์Šคํ…œ์˜ ์ผ์ข…
  • 6. ๋”ฅ๋Ÿฌ๋‹(DL)? ์‹ฌ์ธต ์‹ ๊ฒฝ๋ง(DNN, Deep Neural Network) โ€ข ๋ ˆ์ด์–ด๊ฐ€ ํ•œ ์ธต์œผ๋กœ๋งŒ ๊ตฌ์„ฑ๋œ ๊ฒƒ์ด ์•„๋‹ˆ๋ผ ์—ฌ๋Ÿฌ ์ธต, ๋‹ค์‹œ ๋งํ•ด ๊นŠ์€ ์ธต์œผ๋กœ ๊ตฌ์„ฑ๋œ ์ธ๊ณต ์‹ ๊ฒฝ๋ง ์‹ฌ์ธต ์‹ ๊ฒฝ๋ง์ด ํ•™์Šตํ•˜๋Š” ๊ณผ์ • -> ๋”ฅ๋Ÿฌ๋‹ ใ€Œ ใ€
  • 8. ํ”„๋กœ๊ทธ๋ž˜๋ฐ ๋ฐฉ์‹ & ๋จธ์‹ ๋Ÿฌ๋‹ ๋ฐฉ์‹ ๊ธฐ์กด ์ปดํ“จํ„ฐ ํ”„๋กœ๊ทธ๋žจ ๋จธ์‹  ๋Ÿฌ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜ ์ž…๋ ฅ๊ฐ’์„ ๋ฐ›์œผ๋ฉด ๊ฒฐ๊ณผ๊ฐ’์„ ์ถœ๋ ฅํ•˜๋Š” ๋ฐฉ์‹ ๋‚ด๋ถ€์  ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด ์ค‘์š” ์ž…๋ ฅ๊ฐ’๊ณผ ๊ฒฐ๊ณผ๊ฐ’์„ ๊ฐ€์ง€๊ณ  ์ปดํ“จํ„ฐ๊ฐ€ ํ•™์Šต์„ ํ†ตํ•˜์—ฌ ๊ทœ์น™์„ ๋„์ถœ ํ•ด๋‚ธ ๋‹ค์Œ ์ƒˆ๋กœ์šด ์ž…๋ ฅ๊ฐ’์ด ์ฃผ์–ด์กŒ์„ ๋•Œ ๊ฒฐ๊ณผ๋ฅผ ์˜ˆ์ธก ๋ฐ์ดํ„ฐ๊ฐ€ ์ค‘์š”. ๋ฐ์ดํ„ฐ ์ค‘์‹ฌ & ๋ชจ๋ธ ์ค‘์‹ฌ
  • 10. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ์ง€๋„ํ•™์Šต ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์‹œํ‚ฌ ๋•Œ ๋‹ต๊นŒ์ง€ ๊ฐ™์ด ์•Œ๋ ค์ฃผ๋Š” ๋ฐฉ์‹์˜ ํ•™์Šต ๋ฐฉ๋ฒ• ๊ฐ•์•„์ง€์™€ ๊ณ ์–‘์ด ์‚ฌ์ง„์„ ๊ตฌ๋ถ„ํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ํ•™์Šตํ•  ๋•Œ ์‚ฌ์ง„๊ณผ ํ•จ๊ป˜ ์ด๊ฒƒ์ด ๊ฐ•์ด์ง€์ธ์ง€ ๊ณ ์–‘์ด์ธ์ง€ ์•Œ๋ ค์ฃผ๋Š” ๊ฒƒ
  • 11. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ์ง€๋„ํ•™์Šต โ€ข ๋ ˆ์ด๋ธ”์ด ๋‹ฌ๋ฆฐ ๋ฐ์ดํ„ฐ๋กœ ํ•™์Šต ํ›„ ์ƒˆ๋กœ์šด ๋ฐ์ดํ„ฐ๊ฐ€ ์–ด๋А ๊ทธ๋ฃน์— ์†ํ•˜๋Š”์ง€ ์ฐพ์•„๋‚ด๋Š” ๋ฐฉ๋ฒ• โ€ข ์˜ˆ์ธกํ•˜๋ ค๋Š” ๊ฐ’์ด ์ด์‚ฐ ๊ฐ’ ์ผ ๊ฒฝ์šฐ (์ž๋™์ฐจ, ์‚ฌ๋žŒ, ๋„๋กœ) โ€ข ์˜ˆ๋กœ โ€˜์ŠคํŒธ ๋ฉ”์ผ ๊ตฌ๋ถ„โ€™ โ€ข ์—ฐ์†์ ์ธ ์ˆซ์ž(์‹ค์ˆ˜)๋ฅผ ์˜ˆ์ธกํ•˜๋Š” ๊ฒƒ โ€ข ์˜ˆ์ธกํ•˜๋ ค๋Š” ๊ฐ’์ด ์—ฐ์† ๊ฐ’ ์ผ ๊ฒฝ์šฐ (0.31, 0.301, 0.3001) โ€ข ์˜ˆ๋กœ โ€˜์•„ํŒŒํŠธ๊ฐ€๊ฒฉ ์˜ˆ์ธกโ€™, โ€˜๊ธฐ์˜จ ์˜ˆ์ธกโ€™
  • 12. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์‹œํ‚ฌ ๋•Œ ๋‹ต์„ ์•Œ๋ ค์ฃผ์ง€ ์•Š๊ณ , ๋ฐ์ดํ„ฐ์˜ ํŠน์„ฑ์— ๋”ฐ๋ผ ๊ด€๊ณ„๋‚˜ ํŒจํ„ด์„ ์ฐพ์•„ ์Šค์Šค๋กœ ํ•™์Šตํ•˜๋Š” ๋ฐฉ์‹ ๋™๋ฌผ ์‚ฌ์ง„์„ ๋ถ„๋ฅ˜ํ•  ๊ฒฝ์šฐ ์‚ฌ์ง„์˜ ๊ตฌ์กฐ๋‚˜ ํŠน์„ฑ ๋ณ„๋กœ ๋ถ„๋ฅ˜
  • 13. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต ๊ตฐ์ง‘ํ™” (Clustering) โ€ข ๋ ˆ์ด๋ธ”์ด ์—†๋Š” ํ•™์Šต ๋ฐ์ดํ„ฐ๋“ค์˜ ํŠน์ง•์„ ๋ถ„์„ํ•˜์—ฌ ์œ ์‚ฌํ•œ ํŠน์ง•์„ ๊ฐ€์ง„ ๋ฐ์ดํ„ฐ ๋ผ๋ฆฌ ๊ทธ๋ฃนํ™” ํ•˜๋Š” ๊ฒƒ ๊ตฐ์ง‘ํ™”์˜ ๊ณผ์ • โ€ข ๋ฐ์ดํ„ฐ๋ฅผ ์ขŒํ‘œํ™” -> ๊ฐ€๊นŒ์šด ๊ฒƒ๋“ค ๋ผ๋ฆฌ ๋ชจ์•„ ํ•˜๋‚˜์˜ ๊ทธ๋ฃน์„ ๋งŒ๋“ ๋‹ค ( ์ขŒํ‘œ์ƒ ๊ฐ€๊น๋‹ค๋Š” ๊ฒƒ์€ ๋ฐ์ดํ„ฐ๊ฐ€ ๋น„์Šทํ•˜๋‹ค๋Š” ๊ฒƒ์„ ์˜๋ฏธ)
  • 14. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต ์—ฐ๊ด€ ๊ทœ์น™ (association rule) โ€ข ์„œ๋กœ ์—ฐ๊ด€๋œ ํŠน์ง•์„ ์ฐพ์•„ ๋‚ด๋Š” ๊ฒƒ โ€ข ์ฃผ๋กœ ์ถ”์ฒœ๊ณผ ๊ด€๋ จ๋œ ๋ถ€๋ถ„์— ์‚ฌ์šฉ (์žฅ๋ฐ”๊ตฌ๋‹ˆ ๋ถ„์„) โ€ข ๋ผ๋ฉด์„ ๊ตฌ์ž…ํ•œ ์‚ฌ๋žŒ์€ ๊ณ„๋ž€์„ ๊ตฌ์ž…ํ•  ํ™•๋ฅ ์ด ๋†’๋‹ค โ€ข ์ฆ‰, ๋ผ๋ฉด๊ณผ ๊ณ„๋ž€์€ ์„œ๋กœ ์—ฐ๊ด€์„ฑ(Association)์ด ๋†’๋‹ค ์—ฐ๊ด€์„ฑ์„ ํŒŒ์•… -> ์—ฐ๊ด€์„ฑ์ด ์žˆ๋Š” ์ƒํ’ˆ์„ ์ถ”์ฒœ
  • 15. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๋น„์ง€๋„ํ•™์Šต ๊ตฐ์ง‘ํ™” vs ์—ฐ๊ด€ ๊ทœ์น™
  • 16. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๊ฐ•ํ™”ํ•™์Šต ํ–‰๋™์— ๋Œ€ํ•œ ๋ณด์ƒ์„ ๋ฐ›์œผ๋ฉด์„œ ํ•™์Šตํ•˜์—ฌ ์–ด๋–ค ํ™˜๊ฒฝ ์•ˆ์—์„œ ์„ ํƒ ๊ฐ€๋Šฅํ•œ ํ–‰๋™๋“ค ์ค‘ ๋ณด์ƒ์„ ์ตœ๋Œ€ํ™”ํ•˜๋Š” ํ–‰๋™ ๋˜๋Š” ํ–‰๋™ ์ˆœ์„œ๋ฅผ ์„ ํƒํ•˜๋Š” ๋ฐฉ๋ฒ• ์ฆ‰, ์–ด๋–ค ํ™˜๊ฒฝ ์•ˆ์—์„œ ์ •์˜๋œ ์ฃผ์ฒด(agent)๊ฐ€ ํ˜„์žฌ์˜ ์ƒํƒœ(state)๋ฅผ ๊ด€์ฐฐํ•˜์—ฌ ์„ ํƒํ•  ์ˆ˜ ์žˆ๋Š” ํ–‰๋™(action)๋“ค ์ค‘์—์„œ ๊ฐ€์žฅ ์ตœ๋Œ€์˜ ๋ณด์ƒ(reward)์„ ๊ฐ€์ ธ๋‹ค์ฃผ๋Š” ํ–‰๋™์ด ๋ฌด์—‡์ธ์ง€๋ฅผ ํ•™์Šตํ•˜๋Š” ๊ฒƒ (ex ์•ŒํŒŒ๊ณ , ๊ฐ์ข… ๊ฒŒ์ž„) โ€ข ํ™˜๊ฒฝ(environment) โ€ข ์—์ด์ „ํŠธ(agent) โ€ข ์ƒํƒœ(state) โ€ข ํ–‰๋™(action) โ€ข ๋ณด์ƒ(reward) โ€ข ์ •์ฑ…(policy)
  • 17. ๋จธ์‹ ๋Ÿฌ๋‹์˜ ํ•™์Šต๋ถ„๋ฅ˜ - ๊ฐ•ํ™”ํ•™์Šต ๊ฐ•ํ™”ํ•™์Šต์˜ ์˜ˆ ๋” ๋งŽ์€ ๋ณด์ƒ์„ ๋ฐ›์„ ์ˆ˜ ์žˆ๋Š” ์ •์ฑ…์„ ๋งŒ๋“œ๋Š” ๊ฒƒ์ด ํ•ต์‹ฌ โ€ข ๊ฒŒ์ž„ -> ํ™˜๊ฒฝ(environment) โ€ข ๊ฒŒ์ด๋จธ -> ์—์ด์ „ํŠธ(agent) โ€ข ๊ฒŒ์ž„ํ™”๋ฉด -> ์ƒํƒœ(state) โ€ข ๊ฒŒ์ด๋จธ์˜ ์กฐ์ž‘ -> ํ–‰๋™(action) โ€ข ์ƒ๊ณผ ๋ฒŒ -> ๋ณด์ƒ(reward) โ€ข ๊ฒŒ์ด๋จธ์˜ ํŒ๋‹จ๋ ฅ -> ์ •์ฑ…(policy) 1. ๊ฒŒ์ž„์€ ๊ฒŒ์ด๋จธ์—๊ฒŒ ๊ฒŒ์ž„ํ™”๋ฉด์„ ํ†ตํ•ด ํ˜„์žฌ์˜ ์ƒํƒœ๋ฅผ ๋ณด์—ฌ์ค€๋‹ค. (์บ๋ฆญํ„ฐ์˜ ์œ„์น˜๋‚˜ ์žฅ์• ๋ฌผ์˜ ์œ„์น˜ ๋“ฑ๋“ฑ) 2. ํ˜„์žฌ์˜ ์ ์ˆ˜๋„ ์•Œ๋ ค์ค€ ๋’ค ๊ฒŒ์ด๋จธ์—๊ฒŒ๋Š” ์ ์ˆ˜๊ฐ€ ๋†’์•„์ง€๋Š” ๊ฒƒ์ด ์ƒ์ด๊ณ  ๋ฐ˜๋Œ€๋กœ ์ ์ˆ˜๊ฐ€ ๋‚ฎ์•„์ง€๋Š” ๊ฒƒ์ด ๋ฒŒ์ด๋‹ค. 3. ํ”Œ๋ ˆ์ด(๊ด€์ฐฐ)๋ฅผ ํ•˜๋ฉฐ ์–ด๋–ป๊ฒŒ ํ•˜๋ฉด ๋” ๋งŽ์€ ๋ณด์ƒ์„ ์–ป๊ฒŒ ๋˜๋Š”์ง€, ๋˜ํ•œ ๋” ์ ์€ ๋ฒŒ์„ ๋ฐ›๊ฒŒ ๋˜๋Š”์ง€ ์•Œ๊ฒŒ ๋œ๋‹ค. 4. ๊ทธ๋กœ ์ธํ•ด ๊ฒŒ์ด๋จธ์˜ ํŒ๋‹จ๋ ฅ์ด ๊ฐ•ํ™”๋œ๋‹ค. 5. ๊ฐ•ํ™”๋œ ํŒ๋‹จ๋ ฅ์— ๋”ฐ๋ผ ๊ฒŒ์ž„์„ ์กฐ์ž‘ํ•œ๋‹ค 6. ๊ทธ ์กฐ์ž‘์€ ๊ฒŒ์ž„์— ๋ณ€ํ™”๋ฅผ ๊ฐ€์ ธ์˜จ๋‹ค.
  • 21. MNIST Dataset MNIST Dataset & Fashion MNIST Dataset โ€ข 28ร—28์˜ ํฌ๊ธฐ์˜ 0~9 ์ˆซ์ž ์ด๋ฏธ์ง€ 70,000๊ฐœ MNIST Dataset & Fashion MNIST Dataset โ€ข 28ร—28์˜ ํฌ๊ธฐ์˜ ํ”ฝ์…€ ์ด๋ฏธ์ง€ 70,000๊ฐœ 0 T-shirt/top 1 Trouser 2 Pullover 3 Dress 4 Coat 5 Sandal 6 Shirt 7 Sneaker 8 Bag 9 Ankle boot
  • 22. ๋ฐ์ดํ„ฐ ์˜ˆ์ธก ๋ฐ์ดํ„ฐ ์ค€๋น„ โ†’ ํ•™์Šต โ†’ ๋ฐฐํฌ โ†’ ํ…Œ์ŠคํŠธ โ†’ ์˜ˆ์ธก
  • 24. ์ž๋ฃŒ ์ถœ์ฒ˜ https://blog.lgcns.com/2212 https://hongong.hanbit.co.kr/ai-%EB%AC%B4%EC%97%87%EC%9D%B8%EA%B0%80-%EC%9D%B8%EA%B3%B5%EC%A7%80%EB%8A%A5- %EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D-%EB%94%A5%EB%9F%AC%EB%8B%9D-%EC%B0%A8%EC%9D%B4%EC%A0%90- %EC%B4%9D%EC%A0%95%EB%A6%AC/ https://wikidocs.net/21679 https://codong.tistory.com/37 https://blog.illunex.com/%EB%A8%B8%EC%8B%A0%EB%9F%AC%EB%8B%9D%EA%B8%B0%EA%B3%84%ED%95%99%EC%8A%B5%EC%97%90- %EA%B4%80%ED%95%98%EC%97%AC-1%ED%83%84/ https://opentutorials.org/module/4916/28934 https://school.coding-x.com/lesson/138 https://kimeunh3.github.io/machine%20learning/ml_05/ https://labs.brandi.co.kr/2018/05/17/ohyj.html