Youngwoo (@clover_cs) 's Twitter Profile
Youngwoo

@clover_cs

Postdoc at University of Virginia,
I work on Information Retrieval, Natural Language Processing

ID: 2406920371

calendar_today23-03-2014 12:38:07

29 Tweet

38 Followers

64 Following

Youngwoo (@clover_cs) 's Twitter Profile Photo

Our paper has been accepted to #EMNLP2024 main conference! Our 'relevance thesaurus' reveals neural IR models show bias in prioritizing certain car brands over others. Curious about the implications? Check out our work: arxiv.org/abs/2410.03584

Our paper has been accepted to #EMNLP2024 main conference! Our 'relevance thesaurus' reveals neural IR models show bias in prioritizing certain car brands over others.
Curious about the implications? Check out our work: arxiv.org/abs/2410.03584
Youngwoo (@clover_cs) 's Twitter Profile Photo

NLP Academia ์—์„œ๋Š” Scala๋กœ ์ฝ”๋”ฉํ•˜๋Š”๊ฒŒ ๊ฝค๋‚˜ ๋Œ€์„ธ์ธ๊ฒƒ ๊ฐ™๋‹ค. NLP๋ž‘ Scala ๋ž‘ ๊ถํ•ฉ์ด ์ž˜ ๋งž๋Š”๊ฑธ๊นŒ??

Youngwoo (@clover_cs) 's Twitter Profile Photo

์˜ค๋Š˜ ๋žฉ์—์„œ ICLR ๋…ผ๋ฌธ์„ ๋ฆฌ๋ทฐํ•˜๋Š”๋ฐ, ๊ต์ˆ˜๋‹˜์ด ์ข€ ์ˆ˜์ค€ ๋–จ์–ด์ง€๋Š”๊ฒƒ ๊ฐ™๋‹ค๊ณ  ์ง€์ ํ•˜์…จ๋‹ค. ์—ญ์‹œ ํƒ‘ํ‹ฐ์–ด ์ˆ˜์ค€ ํ•™ํšŒ๋Š” ์•„๋‹Œ๊ฐ€..ใ…‹ใ…‹

Youngwoo (@clover_cs) 's Twitter Profile Photo

์˜์–ด๋ฅผ ๋นจ๋ฆฌ ์ฝ์œผ๋ ค๋ฉด ๋Š์–ด์ฝ๊ธฐ๋ฅผ ์ž˜ํ•ด์•ผ ํ•  ๊ฒƒ ๊ฐ™๋‹ค. ๊ทธ๋ž˜์„œ ๋Š์–ด์ฝ๋Š” ์š”๋ น์„ ์ •๋ฆฌํ•œ ๋งํฌ : facebook.com/bookworm.kr/poโ€ฆ

Youngwoo (@clover_cs) 's Twitter Profile Photo

<์ฒญ์ถ˜์€ ๋งˆ์Œ๊ณผ ๋ชธ์ด ์„ธ์ƒ์˜ ํ•œ๋์—์„œ ๋‹ค๋ฅธ ๋๊นŒ์ง€ ๋™์— ๋ฒˆ์ฉ, ์„œ์— ๋ฒˆ์ฉ ํ•˜๋ฉฐ ๋‚˜๋‹ค๋‹ ๋•Œ๋‹ค. ๋‹ค๋ฅธ ๋‚˜๋ผ์˜ ํ’์†์„ ์ฒดํ—˜ํ•ด ๋ณด๊ณ , ์–ด๋–ค ๋ถ€ํฅํšŒ์—์„œ ๊ฐœ์ข…ํ•˜๋ฉฐ, ํ˜•์ด์ƒํ•™์„ ๋‘๋ฃจ ์„ญ๋ ตํ• ๋•Œ๋‹ค. ์ Š์€ ํ˜ˆ๊ธฐ์˜ ๋ฐฉ์ข…์— ๋Œ€ํ•œ ์˜›๋ง์—๋Š” ์ผ๋ง์˜ ์˜๋ฏธ๊ฐ€ ์žˆ๋‹ค>

Youngwoo (@clover_cs) 's Twitter Profile Photo

์›๊ธฐ ์™•์„ฑํ•œ ์‹œ์ ˆ์˜ ๋ณ‘์„ ์•“์ง€ ์•Š๊ณ  ๊ทธ ์‹œ์ ˆ์„ ์˜์›ํžˆ ๋๋‚ด ๋ฒ„๋ฆฐ ์‚ฌ๋žŒ์€ ์˜ˆ๋ฐฉ ์ ‘์ข…์„ ๋ฐ›์ง€ ์•Š์€ ์•„๊ธฐ์ฒ˜๋Ÿผ ๋ฏฟ์„ ์ˆ˜ ์—†๋Š” ๋Œ€์ƒ์ด๋‹ค. - '๊ฒŒ์œผ๋ฅธ ์ž๋ฅผ ์œ„ํ•œ ๋ณ€๋ช…' ์ค‘์—์„œ

Youngwoo (@clover_cs) 's Twitter Profile Photo

๋…ผ๋ฌธ์œผ๋กœ ์œ ์šฉํ•œ ์ง€์‹์„ ๊ณต์œ ํ•˜์ž!! ์–!!

Youngwoo (@clover_cs) 's Twitter Profile Photo

ํŠธ์œ„ํ„ฐ์—์„œ ์‹ ๊ธฐ์ˆ ์ด ๋‚˜์˜ค๋Š”๊ฑธ ๋ณด๊ณ  ์žˆ์ž๋ฉด ์„ธ์ƒ์€ ์ด๋ ‡๊ฒŒ ๋นจ๋ฆฌ ๋ณ€ํ•˜๋Š”๋ฐ, ๋‚œ ์•„๋Š”๊ฒŒ ์—†์–ด์„œ ๋ถ€๋„๋Ÿฝ๋‹ค ใ… _ใ… 

Youngwoo (@clover_cs) 's Twitter Profile Photo

aclweb.org/anthology/D16-โ€ฆ Conclusion : transfer leanring in NLP has benefits, if they are semantically similar. 1. If you transfer, the output layer must be post-trained for target domain 2. Multi-Task Learning is best, Initializing with trained parameters is ok

Youngwoo (@clover_cs) 's Twitter Profile Photo

arxiv.org/pdf/1705.02364โ€ฆ Conclusion : You can use NLI corpus to use for many other NLP tasks! They said "This model works!"

Youngwoo (@clover_cs) 's Twitter Profile Photo

์„œํ”ผ์Šค ๋žฉํƒ‘์„ ์งˆ๋ €๋‹ค! ๋งฅ์—์–ด๋ณด๋‹ค ๊ฐ€๋ณ๋‹ค. (13.5์ธ์น˜) ์ปดํŒฉํŠธํ•จ์ด ์ •๋ง ๋งˆ์Œ์— ๋“ ๋‹ค.