Taeckyung Lee
@taeckyung_lee
Ph.D. Student @ KAIST (Advisor: Prof. Sung-Ju Lee). Collaborating with Prof. Jinwoo Shin (KAIST) and Prof. Taesik Gong (UNIST).
We're happy to announce that our work on test-time adaptation with binary feedback has been accepted to #ICML2025! We will discuss how we can utilize human feedback for TTA. Special thanks to collaborators @sornswgn @JunsuKim97 @jinwoos0417 @Taesik_MobileAI @wewantsj !!
You can now check our paper! We are hosting a poster session from 11:00 a.m. to 1:30 p.m. on July 17th. Please feel free to reach out to me to connect at ICML in Vancouver. - Arxiv: arxiv.org/abs/2505.18514 - Openreview: openreview.net/forum?id=A5l37…
We're happy to announce that our work on test-time adaptation with binary feedback has been accepted to #ICML2025! We will discuss how we can utilize human feedback for TTA. Special thanks to collaborators @sornswgn @JunsuKim97 @jinwoos0417 @Taesik_MobileAI @wewantsj !!
Entropy minimization is often used to increase the accuracy of models on unlabeled data, but it isn’t clear why it works. In our new ICML paper, we show that it clusters the embeddings of its inputs. With @ziv_ravid, @ylecun, @MatthiasBethge arxiv.org/pdf/2405.05012 1/5 🧵👇
See you at CVPR '24!
📄 Thrilled to announce that our research, "ATTA: Label-Free Accuracy Estimation for Test-Time Adaptation," has been accepted at #CVPR2024! Kudos to the awesome team: @taeckyung_lee, @sornswgn, and @wewantsj 👏 Stay tuned for our camera-ready version and code!
Excited to announce that our research on robust test-time adaptation for noisy streams has been accepted at #NeurIPS2023! Huge thanks to my incredible collaborators: @hai_yewon, @taeckyung_lee, @sornswgn, and @wewantsj. 🙌 Stay tuned for the code and camera-ready version!