Sunnie S. Y. Kim
@sunniesuhyoung
Followers
2K
Following
3K
Media
69
Statuses
907
PhD student @PrincetonCS 🐯 AI + HCI research on responsible and human-centered AI. she/her ☀️ https://t.co/c3atPcWlR1
🇰🇷→Yale→TTIC→Princeton
Joined May 2019
There is a lot of interest in estimating LLMs' uncertainty, but should LLMs express uncertainty to end users? If so, when and how? . In our #FAccT2024 paper, we explore how users perceive and act upon LLMs’ natural language uncertainty expressions. 1/6
8
62
358
Super honored and happy to be selected as an EECS Rising Star and a Siebel Scholar🌟 I'll also be on the job market this year looking for positions where I can continue AI+HCI research on responsible AI!. @PrincetonCS @EPrinceton.
11
14
214
In this difficult time, I'm grateful to share that I'll be starting my PhD at @PrincetonCS this fall. Thank you to everyone who helped me with the application process and the decision🧡.
7
2
163
🤯🤩🤯 to see our work (on reducing biases in CV models) discussed in "Foundations of Computer Vision" . [396]
2
18
146
what's a non-creepy way of telling my co-authors how much I love working with them😳.
12
1
147
Our paper on understanding XAI needs, uses, and perceptions in a real-world human-AI interaction context🐦 has been conditionally accepted to #CHI2023! Co-authored with @watkins_welcome @orussakovsky @ruthcfong @andresmh🧡 Can't wait to share more about it!!.
5
7
134
This coming Monday, June 19th, join us at the 2nd Explainable AI for CV (XAI4CV) Workshop at #CVPR2023! We have an amazing program scheduled 🤩. All materials and info at @CVPR @CVPRConf @XAI_Research @trustworthy_ml
2
31
126
🚨 Announcing the 2nd Explainable AI for CV (XAI4CV) workshop at #CVPR2023. We're excited to bring together researchers and practitioners to discuss the unique challenges and opportunities in XAI for CV and push the frontiers of the field!.
2
28
107
Gave my first research talk at It was at 8am my time, so I set up 10 alarms to make sure I wake up on time and drank lots of ☕ The talk was short (~35min) and a quarter of the audience were friendly faces, but this is still a milestone, right? 😊.
1
2
101
How do end-users trust the AI system they interact with?. In our #FAccT2023 paper, we interviewed 20 end-users of a real-world AI application, the @MerlinBirdID app, and explored what factors affect their trust in AI and how. 1/6
4
15
92
"Help Me Help the AI" received an Honorable Mention Award at #CHI2023 🏅 Huge thanks and congrats to the team @watkins_welcome @orussakovsky @ruthcfong @andresmh 😳🥺🤩🥰🥳.
❓ How can explainable AI (XAI) methods support human-AI interaction in the real world?. We interviewed 20 end-users of a real-world AI application, the @MerlinBirdID app, and asked about their XAI needs, uses, and perceptions. 1/7.
5
11
91
📢 Recruiting for a paid study about Explainable AI and birds identification 🐦. We're looking for US-based participants for a 1.5 hour study on Zoom (payment: $25). Open to diverse levels of ML and birding experience. If interested, please fill out 👇
4
21
86
Had fun attending the NYC Computer Vision Day and giving a talk on "Bridging Computer Vision and HCI" 🤝. Event: My research: Hope to see more CV+HCI work!! 💛
1
3
80
I was awarded the NSF Graduate Research Fellowship!! I feel lucky mostly, but also encouraged and happy. Thanks to my mentors and friends for helping with my application materials and supporting me in this journey💖. @YaleDataScience @TTIC_Connect @PrincetonCS @VisualAILab.
3
6
77
Made it to Chicago for my first FAccT! . I'd love to meet up and chat about transparency, explainability, and trust in AI (especially LLMs and CV models). I'll be around the whole conference and definitely at paper session 1 tmrw (6/12 Monday 3:15-4:15pm in W196BC) to present 👇.
How do end-users trust the AI system they interact with?. In our #FAccT2023 paper, we interviewed 20 end-users of a real-world AI application, the @MerlinBirdID app, and explored what factors affect their trust in AI and how. 1/6
2
4
74
📢 Recruiting participants for a paid study about LLMs. I'm looking for US-based adults for a 1 hour study on Zoom (payment: $15 Amazon gift card), especially those with technical knowledge of LLMs. Sign up at Please RT and help out a grad student 😁.
1
37
74
Thank you #CHI2023 community for being so welcoming, kind, and inspiring. I learned so much and am so heartened🥺❤️ Couldn't have asked for a better first CHI!
6
1
70
❓ How can explainable AI (XAI) methods support human-AI interaction in the real world?. We interviewed 20 end-users of a real-world AI application, the @MerlinBirdID app, and asked about their XAI needs, uses, and perceptions. 1/7.
Our paper on understanding XAI needs, uses, and perceptions in a real-world human-AI interaction context🐦 has been conditionally accepted to #CHI2023! Co-authored with @watkins_welcome @orussakovsky @ruthcfong @andresmh🧡 Can't wait to share more about it!!.
2
14
66
Thank you for the excellent workshop and for giving us the opportunity to talk about our work! It's really valuable to discuss HCI papers at CV/ML conferences and vice versa. Cheers to cross-pollination of perspectives 🐝🌼. @eccvconf #ECCV2024 @XAI_Research
Less than three days to go for the eXCV Workshop at #ECCV2024! Join us on Sunday from 14:00-18:00 in Brown 1 to hear about the state of XAI research from an exciting lineup of speakers!. @orussakovsky, @vidal_rene, @sunniesuhyoung, @YGandelsman, @zeynepakata . @eccvconf . (1/4)
0
5
65
Considered using concept-based explanation methods (e.g., NetDissect, TCAV, Concept Bottleneck)?. Check out our paper (to appear in #CVPR2023 🎉) to maximize explanation quality!. By Vikram V. Ramaswamy @sunniesuhyoung @ruthcfong @orussakovsky.
0
9
58
#CHI2024 was 💚🌴🐢 Special thanks to the Doctoral Consortium mentors & fellow participants and friends who adopted me into their groups. I had a great time because of your kindness. Hope to be back soon!!
0
0
57
Apply apply! I had a great time at this year's Doctoral Consortium. It was really wholesome and inspiring 🌱 Happy to chat about my experience and answer any questions. This was my application:
Ph.D. student roughly midway through your thesis work? Seeking additional mentoring and peer-support? 🤝 .Apply to the Doctoral Consortium at CHI 2025. It's a great opportunity to network and gain feedback to your research. 📚 .Apply for in-person DC by October 24, 2024 (AoE)
0
8
55
XAI4CV is back at #CVPR again with a fantastic program 🤩 Join us on Tuesday, June 18th in Arch 2A!. All info at @tmiller_uq @suinleelab @kerstingAIML @sukrutrao @KolekDe @LenkaTetkova @_Kate_Morrison_ @deeptigp . #CVPR2024 @CVPR @CVPRConf @XAI_Research
This coming Monday, June 19th, join us at the 2nd Explainable AI for CV (XAI4CV) Workshop at #CVPR2023! We have an amazing program scheduled 🤩. All materials and info at @CVPR @CVPRConf @XAI_Research @trustworthy_ml
1
16
54
I started co-managing the @XAI_Research twitter this week! Excited to help grow and foster the #XAI community🌱 together with @NickKroeger1 (who has been spearheading everything) and others. Hoping it grows as much as @ml_collective @trustworthy_ml that I admire❤️.
2
8
53
Completed the #IBMQuantumChallenge!! . Should I try beating 45 (currently 46, 2nd place) by picking more of @kyungjoo_noh's brains?
3
3
52
Huge thanks to all speakers, presenters, attendees, and organizers 🙌 We hope you had fun at the XAI4CV workshop!. Please let us know what went well, what could be improved, and if you’re interested in organizing next year: @CVPR @CVPRConf #CVPR2023
This coming Monday, June 19th, join us at the 2nd Explainable AI for CV (XAI4CV) Workshop at #CVPR2023! We have an amazing program scheduled 🤩. All materials and info at @CVPR @CVPRConf @XAI_Research @trustworthy_ml
1
8
51
2 papers accepted to #CVPR2021! Preprints are available for both (see below for links and threads) 🥳.
3
1
50
What a pleasant surprise! Thanks for covering our work 🙈. "Help Me Help the AI": Understanding How Explainability Can Support Human-AI Interaction @sunniesuhyoung @watkins_welcome @orussakovsky @ruthcfong @andresmh. Learn more 👉
Just published CHI 2023 Editors' Choice on "Human-Centered AI". A few of our editors identified CHI papers that really stood out to them because they were interesting and made a significant contribution to human-centered AI.
0
7
43
My first work with @PrincetonCS @VisualAILab is now on arXiv! ✨. "Fair Attribute Classification through Latent Space De-biasing" by Vikram V. Ramaswamy, me, and Olga Russakovsky. Project page: Paper: Code:
New work "Fair Attribute Classification through Latent Space De-biasing" is out on arxiv. By Vikram Ramaswamy, @sunniesuhyoung, @orussakovsky. for paper and code.
1
2
45
I attended the #HCXAI workshop twice and loved it so much (it really transformed how I think about XAI) ❤️ that I'm helping out with the organization this year at #CHI2024. Please spread the word and come join us! . @XAI_Research @trustworthy_ml @ml_collective.
Why should you submit to the Human-centered Explainable AI workshop (#HCXAI) at #chi2024?. Come for your love for amazing XAI research; stay for our supportive community. That's what 300+ attendees from 18+ countries have done. Here's a snippet of what they think ⤵️. 🧵1/n
0
6
43
In today's lab meeting, my advisor & lab members gave me a toast for submitting my first paper with @VisualAILab! We did it with coffee, water, etc. over zoom (traditionally it's done with 🍾 after the deadline ~3am). It was the most lovely thing🥺💕.
0
1
43
🥳 to share the accepted papers to the #CVPR2023 XAI4CV workshop proceedings track! Here's a sneak peek👀. The non-proceedings track will receive "extended abstract" submissions until May 19. Consider submitting to share your work with the XAI4CV community!
🚨 Announcing the 2nd Explainable AI for CV (XAI4CV) workshop at #CVPR2023. We're excited to bring together researchers and practitioners to discuss the unique challenges and opportunities in XAI for CV and push the frontiers of the field!.
1
6
42
Thanks for highlighting our upcoming #FAccT2024 paper (that examines how LLMs' uncertainty expression affects user reliance and trust) as good reads in human-centered AI 😊. 📝 🧵 w/ @QVeraLiao @mihaela_v @ballard4_4 @jennwvaughan.
📢Happy to share that I'm now an editor for the Medium-based "Human-Centered AI" Publication! 😄.With Justin Weisz, @wernergeyer, @watkins_welcome, @qiaosi_wang, @alansaid. Check out our intros here and consider submitting your article ideas:
0
4
40
The XAI4CV workshop will be back at #CVPR2024 led by the amazing @indu_panigrahi!. This year the organizing team spans 8 institutions @sukrutrao @KolekDe @LenkaTetkova @_Kate_Morrison_ @deeptigp + Jawad, Pushkar, Vikram ❤️. @CVPR @CVPRConf @XAI_Research.
Huge thanks to all speakers, presenters, attendees, and organizers 🙌 We hope you had fun at the XAI4CV workshop!. Please let us know what went well, what could be improved, and if you’re interested in organizing next year: @CVPR @CVPRConf #CVPR2023
0
11
39
Growing up in Korea, no one including myself expected me to go into STEM, esp. not CS. I was fortunate to have supportive mentors and attend university in the US where faculty diversity and awareness of lack thereof are slightly better (otherwise I don't think I would have been.
""".The 30 women faculty in [KAIST] . thought that highlighting the research excellence of women faculty can have a profound impact . in Korean academia [and] was immediately implemented through Breakthroughs . a bi-annual webzine.""". by @aliceoh.
2
3
40
Had a lovely first @FAccTConference! Caught up with friends, made new ones, met people whose work I admire and have built on, and added many many papers to my reading list. Hope to be back soon!. Now off to Vancouver for #CVPR2023 😴🤩.
0
2
39
Truly enjoyed this interdisciplinary (CS + Psych + Philosophy) workshop today at Princeton as someone doing interdisciplinary research on human understanding of AI systems. Thank you @TaniaLombrozo and Stephen Grimm for organizing and the speakers for the amazing discussion!
0
1
37
@tomgoldsteincs Here's a summary of post-hoc XAI methods for computer vision models. Example-based: Representer points (, Influence functions (. Counterfactual: (1/2).
1
4
35
XAI4CV workshop will be back at #CVPR2023 🥳 Stay tuned!. Organized by @filipradenovic @deeptigp @dubeya_ @ruthcfong Vikram Ramaswamy & myself. @CVPR @XAI_Research @trustworthy_ml.
Full house plus additional ~80 Zoom attendees today at our XAI4CV @CVPR workshop. Thank you to all who participated, and I hope you enjoyed it.
0
7
34
🧵 of various exciting things I'm seeing and learning at #CVPR2022. keywords: interpretability/XAI, model steering, fairness, responsible/human-centered AI.
1
4
35
First in-person #CVPR2022! Having started grad school during the pandemic, everything feels unreal. but looking forward to meeting people, learning a bunch of stuff, and exploring New Orleans.
2
0
33
Can XAI explanations help human users distinguish between correct and incorrect model predictions? . Check out our #ECCV2022 paper 🐝HIVE🍯 on interpretability evaluation!. Preprint, code, and short talk videos are available at
HIVE: Evaluating the Human Interpretability of Visual Explanations ( accepted to #ECCV2022 -- by @sunniesuhyoung, @nicole__meister, Vikram V. Ramaswamy, @ruthcfong and @orussakovsky.
1
3
32
yay recognized as an outstanding reviewer (2 years in a row✌️) @repro_challenge is an amazing initiative so glad to have contributed!.
Thanks for your patience : the final camera ready process took some time. We are now excited to announce the decisions of MLRC 2022 on our website: We are also happy to announce the outstanding papers and reviewers! Congratulations to all!.
1
0
30
#CVPR2023 workshop proceedings are out! See for the 17 🔥 papers accepted to the XAI4CV workshop.
🥳 to share the accepted papers to the #CVPR2023 XAI4CV workshop proceedings track! Here's a sneak peek👀. The non-proceedings track will receive "extended abstract" submissions until May 19. Consider submitting to share your work with the XAI4CV community!
0
4
30
XAI meetups at #CVPR2022!. 🍕 after the #XAI4CV workshop on Monday & Yesterday's lunch organized by @giangnguyen2412 and me. They were super fun, let's do more of these in the future😊
1
7
30
Hi #FAccT2024! Join the trust and reliance session at Wed 11:35am in the Sao Conrado room to hear from me ☀️, Naina Balepur, @Arianna_Manzini, @RuotongWang, @ZiyangGuo27. I'll talk about LLMs' uncertainty expression and its impact on (over)reliance/trust.
There is a lot of interest in estimating LLMs' uncertainty, but should LLMs express uncertainty to end users? If so, when and how? . In our #FAccT2024 paper, we explore how users perceive and act upon LLMs’ natural language uncertainty expressions. 1/6
1
1
28
Curious what factors affect people's trust in AI and how? (Spoiler: There are many, we organize them in a nice table for you🤓). Thanks @mtlaiethics @atg_abhishek for featuring our #FAccT2023 paper!.
2
8
25
#CVPR2022 just wrapped and I couldn't have asked for a better first in-person CVPR experience!. Gave a workshop talk 🎤.Had inspiring discussions 💡.Caught up with friends ❤.Made a ton of new friends 🤩.Enjoyed New Orleans 🎺.Became part of a community 🥺
1
0
28
That's a wrap! Thank you to everyone who made XAI4CV at #CVPR2024 possible 🎉. What went well and what could be improved? Are you interested in organizing/reviewing for a future workshop or have speaker nominations? . Let us know at @CVPR @CVPRConf.
XAI4CV is back at #CVPR again with a fantastic program 🤩 Join us on Tuesday, June 18th in Arch 2A!. All info at @tmiller_uq @suinleelab @kerstingAIML @sukrutrao @KolekDe @LenkaTetkova @_Kate_Morrison_ @deeptigp . #CVPR2024 @CVPR @CVPRConf @XAI_Research
0
1
27
@tomgoldsteincs Concept-based: TCAV (, NetDissect (, IBD (. I'm personally a fan of this approach and recently wrote a more retrospective/analysis-focused paper on it w/ my coauthors (. (2/2).
1
3
25
1. OMG!!!!!!!!!!! 🥳❤.2. FYI Youn Yuh-Jung is one of the most famous and respected actors in S. Korea🇰🇷.3. Brad Pitt screwing up her name (which she handled gracefully) reminded me of my ugrad commencement when my name was messed up lol. Let's plz learn and practice names😒.
Best Supporting Actress Winner Yuh-Jung Youn's acceptance speech was hysterical. Watch the full thing: #Oscars
0
1
24
Mathematics for Machine Learning by @mpd37 @AnalogAldo @ChengSoonOng. @MIT The Missing Semester of Your CS Education . Always open to suggestions though!. (4/).
4
0
22
this week I got a surprise email from my ugrad advisor and it made me really truly happy😊 got me thinking that I should send (hello how are u) emails more often to friends and previous collaborators, mentors, and profs!.
0
0
23
Excited for #CVPR2021! I'll be presenting at the main conference (2 papers) and 3 workshops: @WiCVworkshop (talk & poster), Responsible CV (talk) and @FGVCworkshop (poster). Also check out all the other things our lab is doing🤩.
Come meet us at #CVPR2021! (1) Vikram Ramaswamy, Sunnie Kim (@sunniesuhyoung) and Olga Russakovsky (@orussakovsky) will be presenting a poster on GAN-based debiasing ( at Session 7, Wed 11am-1:30pm ET.
0
0
22
Below are some freely available textbooks/resources at the top of my list:. Computer Vision: Algorithms and Applications by Richard Szeliski. Fairness and Machine Learning by @s010n @mrtz @random_walker. (3/).
1
0
22
Academic 💕 language (conference version):. "I read/cited/loved your paper". "Look forward to reading your paper". "Let me know when your paper comes out". 😂.
0
1
21
#CVPR2023 XAI4CV workshop proceedings track deadline is 2 weeks away! Submit to share and discuss your work with the community💛. For details see @CVPR @XAI_Research @trustworthy_ml
🚨 Announcing the 2nd Explainable AI for CV (XAI4CV) workshop at #CVPR2023. We're excited to bring together researchers and practitioners to discuss the unique challenges and opportunities in XAI for CV and push the frontiers of the field!.
0
7
21
Just submitted my last course project for the semester. I survived G1!!!!!!!.
0
0
20
Excited to present (my first vision paper🥳) "Deformable Style Transfer" at #ECCV2020! DST is a one-shot, geometry-aware style transfer method that works on non-face images. Work with Nick Kolkin, Jason @Salavon and Greg Shakhnarovich @TTIC_Connect @UChicago
1
5
20
🎤 We have a great line-up of speakers who are experts in the fields of Computer Vision, Machine Learning, Human-Computer Interaction, Visualization, and AI Ethics:. @QVeraLiao @mohitban47 @Marina_MCV @arvindsatya1 @alicexiang @davidbau
1
4
20
I was super excited about my first citation until I realized my name is messed up😥 Why does this happen and what can be done? #phdchat #AcademicTwitter
2
1
18
Princeton AI4ALL (@ai4allorg) started this week! Super excited to introduce students (rising 11th graders😳 I didn't know what AI was until like junior year of college!) to the world of AI⚡.
0
2
19
Do you use or know someone who uses @MerlinBirdID? 🐦🦜🦚 We're looking for Merlin Photo/Sound ID users for our XAI research study!. If you are interested in doing a paid ($30) 1 hour zoom interview, please fill out (also has more info about the study).
1
9
18
Interested in submitting to the #CHI2024 HCXAI workshop ( but not sure where to start?. We're hosting pre-submission office hours next Friday, Feb 2nd, 11am-noon ET! See Upol's post for details 👇.
🎯 Submitting to a CHI workshop can be daunting, esp for 1st timers or under-represented groups. But fear not!. We, at the Human-centered XAI workshop, will host Office hours to help you! 🎉. Limited spots. RSVP ASAP: Pls repost + help us beat the filter.
1
6
18
I'll start grad school in 3 weeks but it feels unreal. is it just me? Can't believe it's already August, I feel like I'm still stuck in March #phdchat #AcademicChatter.
0
0
18
Thanks so much Sara (@sarameghanbeery) for the amazing talk! Our project team in Princeton AI4ALL (@ai4allorg) is focusing on biodiversity monitoring using computer vision, inspired by your work, so it was super cool to hear more about them directly from you. Students loved it🧡
0
0
18
Presented my work at a conference for the first time, and I feel really good about it☺ I was nervous that no one would come to the poster sessions, but I ended up talking to ~10 people. I also feel more like I'm part of the community. Looking forward to many more of these!.
Hello #ECCV2020! My coauthors and I will be available to discuss our paper at the following poster/live q&a sessions:. 8/23 Sunday @WiCVworkshop.6:40-7:55am (PT) / 2:40-3:55pm (BST). 8/25 Tuesday @ECCVconf.1️⃣ 6-8am (PT) / 2-4pm (BST).2️⃣ 4-6pm (PT) / Wed 12-2am (BST).
0
0
17
This work opened my eyes to human-centered explainable AI and shaped how I think about the space. So excited to hear the talk!.
Care about Explainable AI? . Want to learn about how people with and without an AI background interpret AI explanations? Curious about a new type of harmful effect in XAI? . Where : 313B .When: 12-12:15 pm . You'll get exclusive access to behind the scene details on how this
0
1
16
So really taking the time to build a foundation in your chosen area, rather than trying to stay on top of arXiv papers (you can't anyways😂). While this is crucial in the long run, it's hard to do when you don't have long blocks of time. ☀️ might be a perfect time for some!. (2/).
2
0
14
My first computer vision/graphics project and arXiv submission🎉 It was super fun thinking about image feature statistics and definitions of artistic content/style with Nick, @Salavon and Greg!.
1
2
14
Just read a super thorough and thoughtful review of my work🤯😭. Such reviews are so rare, so valuable these days. .
1
0
15
After watching #QueensGambit, @kyungjoo_noh and I've been trying to beat Beth Harmon in @chesscom but we keep losing to Beth at age 10🤣 So instead we're watching the amazing @Anna_Chess winning easily against Beth!
0
0
15
I shriek every time the graduate school calls me an "AI" for assistant in instruction. Why not a TA or something less 🤪?.
0
0
15
Due to requests, we have extended the proceedings track deadline from March 14 (Tuesday) to March 17 (Friday) 5pm PT. Looking forward to the submissions!. #CVPR2023 XAI4CV workshop. @CVPR @XAI_Research @trustworthy_ml.
🚨 Announcing the 2nd Explainable AI for CV (XAI4CV) workshop at #CVPR2023. We're excited to bring together researchers and practitioners to discuss the unique challenges and opportunities in XAI for CV and push the frontiers of the field!.
0
4
15
Whohoo our ML Reproducibility Challenge @repro_challenge submission has been selected for publication in the @ReScienceEds journal!. Work with Sharon Zhang (@sxzhang25), Nicole Meister, Olga Russakovsky (@orussakovsky) @PrincetonCS @VisualAILab.
1
2
15
Miss you already 😭 but so excited for you and all the things you will do!!!! ❤️🎉🙌.
Excited to share I’ll be joining be joining as an Assistant Professor at @CornellInfoSci @Cornell_Tech in Summer 2025! This coming year I’ll be a postdoc at @StanfordHAI with @SanmiKoyejo and Daniel Ho 🎈. I am so grateful to all of my mentors, friends, family. Come visit!
1
0
13
Today in our biweekly Bias in AI reading group ( I led the discussion on two papers that talk about the costs and risks of large language models:.
1
0
13
1
0
13
reached a point in the semester where my friend (@vivienngyn) and I are chatting in overleaf 🤓.
0
0
13
I was one of ~6 undergrad majors in the yale statistics dept in 2017. Over the years, I saw the dept change its name to "statistics and data science" and attract 100+ undergrads including Nathan Chen🏅 What an achievement!!.
TIL Nathan Chen is majoring in Statistics and Data Science. Here he is celebrating getting his R code to work on the first try.
0
0
13
4 days left until the submission deadline for the #HCXAI (Human-Centered Explainable AI) workshop at #CHI2024! Looking forward to the submissions🤩. @acm_chi @XAI_Research @trustworthy_ml.
**DEADLINE EXTENDED** You asked. We listened!. New deadline: Feb 21 for the Human-centered Explainable AI (#HCXAI) workshop at #CHI2024. w/ @mark_riedl @watkins_welcome @CarinaManger @philwintersberg @AndreasRiener @sunniesuhyoung @nielsvberkel . (weblink in ALT text)
0
6
13
"Information-Theoretic Segmentation by Inpainting Error Maximization" by @PedroSavarese, Sunnie S. Y. Kim, Michael Maire, Greg Shakhnarovich, @McAllesterDavid @TTIC_Connect @UChicagoCS #CVPR2021. Paper: Thread by @_onionesque:
"Information-Theoretic Segmentation by Inpainting Error Maximization".proposes a simple and cheap method for unsupervised segmentation that outperforms current state of the art methods -- which are usually expensive and rely on heavy learning machinery
1
1
12
Had a great first day of my first conference @WiMLworkshop! It was inspiring to see and meet so many amazing women in ML💕 #WiML2019 #NeurIPS2019
0
2
12
We find that expressing uncertainty in natural language can be an effective way to reduce overreliance and (over)trust in LLMs. Uncertainty expression led to more cautious behaviors, from taking longer to arrive at an answer to reporting more reliance on outside sources. 3/6.
1
1
12
I woke up at 5am today ☕ for @WiCVworkshop and it was totally worth it! Thank you for putting together a great workshop. I also want to thank @DeviParikh, @IlkeDemir, Caroline Pantofaru, and fellow mentees for the wonderful mentoring session!.
1
1
12
Our results suggest teams building and deploying LLMs and LLM-infused applications should evaluate approaches to mitigate overreliance, including language choices, carefully with users before release. We hope our work serves as an example and guide for such evaluations. 5/6.
1
0
11
@kyungjoo_noh Completed 3 out of 4 exercises! It kind of felt like solving a pset LOL but it was fun😎. #IBMQuantumChallenge
0
1
11
"Fair Attribute Classification through Latent Space De-biasing" by Vikram V. Ramaswamy, Sunnie S. Y. Kim, Olga Russakovsky @PrincetonCS @VisualAILab #CVPR2021. Project page: Thread:
My first work with @PrincetonCS @VisualAILab is now on arXiv! ✨. "Fair Attribute Classification through Latent Space De-biasing" by Vikram V. Ramaswamy, me, and Olga Russakovsky. Project page: Paper: Code:
1
1
11