Changho Shin
@Changho_Shin_
Followers
428
Following
666
Media
3
Statuses
146
Ph.D student at @WisconsinCS @UWMadison
Madison, WI
Joined March 2022
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#ปิ่นภักดิ์EP10
• 684728 Tweets
TLP THE SECRET OF RING
• 681045 Tweets
凱旋門賞
• 130655 Tweets
Villa
• 124871 Tweets
Forest
• 109034 Tweets
HUNTER×HUNTER
• 103814 Tweets
Jets
• 88234 Tweets
Evans
• 77438 Tweets
シンエンペラー
• 58775 Tweets
Falling Up
• 56607 Tweets
Lewandowski
• 52468 Tweets
宇宙戦艦ヤマト
• 50341 Tweets
#TheStarChapter
• 48340 Tweets
Vikings
• 47170 Tweets
オリックス
• 46787 Tweets
中嶋監督
• 46655 Tweets
Rodgers
• 42350 Tweets
Enzo
• 36447 Tweets
Raphinha
• 28673 Tweets
Madueke
• 26561 Tweets
ITZY IMAGINARY TRACK
• 26035 Tweets
ブルーストッキング
• 25465 Tweets
Saleh
• 19571 Tweets
アルリファー
• 19033 Tweets
ロスアンゼルス
• 16954 Tweets
ハンターハンター
• 15500 Tweets
アヴァンチュール
• 14439 Tweets
Darnold
• 13478 Tweets
#乃木坂工事中
• 13066 Tweets
Nkunku
• 10510 Tweets
Jefferson
• 10340 Tweets
Heading to
#NeurIPS
2023 from Dec 10-17!
My collaborators and I will present
1. Fairness in weak supervision
2. How to use LLMs for foundation model test-time adaptation @ R0-FoMo Workshop
3. Improving data wrangling LLMs with better demonstrations @ TRL Workshop
🧵
1
9
32
Curious if you can robustify💪 foundation models🤖 almost for free?(!!)💸 Join us for the poster presentation on “Zero-Shot Robustification of Zero-Shot Models” with
@dyhadila
,
@CaiLinrong
@fredsala
!
1
4
22
🚀 Excited to share our latest research at
#ICML2024
Workshops!
1️⃣ Weak-to-Strong Generalization Through the Data-Centric Lens
What aspects of data induce weak-to-strong generalization? We show that overlap density (the proportion of data with both easy and hard features)
1
6
20
Excited to share our work on "Mitigating Source Bias for Fairer Weak Supervision" with
@SonNicCr
,
@dyhadila
, and
@fredsala
at poster session 1 today! 🚀
Come and let's chat!
1. Mitigating Source Bias for Fairer Weak Supervision: Weak supervision is great for efficiently building labeled datasets—but it can amplify bias. What can we do about this?
1
0
4
0
2
13
While you’re waiting for NeurIPS decisions—check out a fun problem! Can we improve large pre-trained models’ robustness *without* getting more data and fine-tuning?
1
12
42
1
0
5
1. Mitigating Source Bias for Fairer Weak Supervision: Weak supervision is great for efficiently building labeled datasets—but it can amplify bias. What can we do about this?
1
0
4
It turns out, some good things in life are free. We propose a simple method to improve both fairness *and* accuracy in weak supervision. Check it out at Great Hall & Hall B1+B2 (level 1)
#2014
10:45, Tuesday
Paper:
Code:
1
0
4
Feel free to reach out if you wanna chat! Happy to chat about weak supervision, foundation models, optimal transport or anything else!
0
0
2
🔼Check it out at Room 235 - 236, Spotlight talk (TBD), Poster (10:20, 15:20), Friday
1
0
2
2. Foundation Models can Robustify Themselves, For Free (w/
@dyhadila
): How do we make foundation models more robust? We ask them questions about what knowledge they were supposed to use---and use their answers to improve predictions.
Join us at Ballroom A/B, Oral (16:10), Friday
2
0
2
Check it out at Room 235 - 236, Spotlight talk (TBD), Poster (10:20, 15:20), Friday
0
0
1
3. Improving Data-wrangling LLMs via better in-context examples (w/ Joon Suk Huh): Narayan et al. (VLDB 2023) showed LLMs can be used for data wrangling tasks. Our study explores if better demonstrations (quantity, relevancy, diversity) can further improve LLMs in data wrangling.
1
0
1