๋ณธ๋ฌธ ๋ฐ”๋กœ๊ฐ€๊ธฐ

KAIST MASTER๐Ÿ“š/CS231n4

[CS231n] Lecture 5 - Convolutional Neural Networks www.youtube.com/watch?v=bNb2fEVKeEo&list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv&index=5 Stanford University์—์„œ 2017๋…„๋„์— ๊ฐ•์˜ํ•œ CS231n๋ฅผ ๋“ค์œผ๋ฉฐ ์ •๋ฆฌ, ์š”์•ฝํ–ˆ๋‹ค. 4๊ฐ• Convolutional Neural Networks for Visual Recognition ๊ฐ•์˜ ์š”์•ฝ ์‹œ์ž‘! intermediate feature ์„ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค. ๊ฐ„๋žตํ•œ ์—ญ์‚ฌ 1957 : ํผ์…‰ํŠธ๋ก  2012 : Alexnet Convnet์€ ๋น ๋ฅด๊ฒŒ ๋ฐœ์ „ํ•˜์—ฌ classification, detection, segmentation, image captioning ๋“ฑ์— ์ด์šฉ๋œ๋‹ค Fully Connected Layer input์„ ์ญ‰ ํŽด์„œ ํ–‰๋ ฌ๊ณฑํ•จ Con.. 2021. 2. 3.
[CS231n] Lecture 4 - Introduction to Neural Networks www.youtube.com/watch?v=d14TUNcbn1k&list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv&index=4 Stanford University์—์„œ 2017๋…„๋„์— ๊ฐ•์˜ํ•œ CS231n๋ฅผ ๋“ค์œผ๋ฉฐ ์ •๋ฆฌ, ์š”์•ฝํ–ˆ๋‹ค. 4๊ฐ• Convolutional Neural Networks for Visual Recognition ๊ฐ•์˜ ์š”์•ฝ ์‹œ์ž‘! ์ €๋ฒˆ ์‹œ๊ฐ„ ๋ณต์Šต optimization : ๊ฐ€์žฅ ๋‚ฎ์€ ์  ์ฐพ๊ธฐ gradient descent ์ด๋ฒˆ ์‹œ๊ฐ„์—๋Š” analytic gradient๋ฅผ ์–ด๋–ป๊ฒŒ ๊ตฌํ• ์ง€ ๋ฐฐ์šด๋‹ค L์—๋Š” R๋„ ํฌํ•จ๋˜์–ด ์žˆ๋‹ค (regulization) Neural Turing Machine-> ๋งค์šฐ ๋ณต์žก ๊ทธ๋ž˜์„œ backpropagation์ด ๋ญ”๋ฐ? ๋งŒ์•ฝ input์ด ์ €๋ ‡๊ฒŒ ์ฃผ์–ด.. 2021. 2. 1.
[CS231n] Lecture 3 - Loss Functions and Optimization www.youtube.com/watch?v=h7iBpEHGVNc&list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv&index=3 Stanford University์—์„œ 2017๋…„๋„์— ๊ฐ•์˜ํ•œ CS231n๋ฅผ ๋“ค์œผ๋ฉฐ ์ •๋ฆฌ, ์š”์•ฝํ–ˆ๋‹ค. 3๊ฐ• Loss Functions and Optimization ๊ฐ•์˜ ์š”์•ฝ ์‹œ์ž‘! Lecture 2 ๋ณต์Šต ๊ณ ์–‘์ด ์‚ฌ์ง„์„ ์ธ์‹ํ•˜๊ฒŒ ํž˜๋“ค๊ฒŒ ๋งŒ๋“œ๋Š” ์›์ธ ์ง€๋‚œ ์‹œ๊ฐ„์˜ ๋ณต์Šต W์˜ row๋ฅผ visualize ์–ด๋–ค W๊ฐ€ ์ข‹์€ W๋ผ๊ณ  ํ•  ์ˆ˜ ์žˆ์„๊นŒ? car์€ ์˜ˆ์ธก์„ ์ž˜ํ•˜๊ณ  frog๋Š” ์—„์ฒญ ๋ชปํ•จ ์–ด๋–ค W๊ฐ€ best์ธ์ง€ ์•Œ๋ ค๋ฉด loss function์„ ์ด์šฉํ•ด์•ผ ํ•จ ์–ด๋–ค W ๊ฐ€ least bed ํ•œ๊ฐ€? loss function x๋Š” input (ํ”ฝ์…€ ์ด๋ฏธ์ง€) y๋Š” ๋ผ๋ฒจ, ํƒ€๊ฒŸ.. 2021. 1. 31.
[CS231n] Lecture 2 - Image Classification www.youtube.com/watch?v=OoUX-nOEjG0&list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv&index=2 Stanford University ์—์„œ 2017๋…„๋„์— ๊ฐ•์˜ํ•œ CS231n๋ฅผ ๋“ค์œผ๋ฉฐ ์ •๋ฆฌ, ์š”์•ฝํ–ˆ๋‹ค. 2๊ฐ• Image Classification ๊ฐ•์˜ ์š”์•ฝ ์‹œ์ž‘! ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜์˜ ๋ฌธ์ œ์  ์ปดํ“จํ„ฐ๊ฐ€ ๋ณด๋Š” ๊ณ ์–‘์ด! x, y, rgb ์ฐจ์› (800 * 600 * 3) ์กฐ๊ธˆ๋งŒ ์นด๋ฉ”๋ผ ๊ฐ๋„๋ฅผ ๋ฐ”๊พธ๊ฑฐ๋‚˜, ๋น›์˜ ์œ„์น˜๊ฐ€ ๋‹ฌ๋ผ์ง€๊ฑฐ๋‚˜, ๊ณ ์–‘์ด๊ฐ€ ์‡ผํŒŒ ๋ฐ‘์— ์ˆจ์–ด์žˆ๋‹ค๋ฉด ์ด ํ”ฝ์…€ ๊ฐ’๋“ค์€ ํฌ๊ฒŒ ๋ฐ”๋€” ๊ฒƒ์ด๋‹ค. ๋ฐฐ๊ฒฝ์ด ๋ณต์žกํ•˜๊ฑฐ๋‚˜ ์—ฌ๋Ÿฌ ๋งˆ๋ฆฌ๊ฐ€ ๋ญ‰์ณ์žˆ๋Š” ๊ฒฝ์šฐ์—๋„ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ํ˜ผ๋ž€์„ ๊ฒช์„ ๊ฒƒ์ด๋‹ค. ํ•˜์ง€๋งŒ ์‚ฌ๋žŒ์€ ms๋งŒ์— ์‚ฌ์ง„์ด ๊ณ ์–‘์ด๋ผ๋Š” ๊ฒƒ์„ ๊ฐ๋ณ„ํ•ด๋‚ธ๋‹ค. sort number์™€ ๊ฐ™์€ ๊ฒƒ์€ ๋ช…ํ™•ํ•œ.. 2021. 1. 30.