interpretability analysis
Sparse autoencoders reveal selective remapping of visual concepts during adaptation
Hyesu Lim, Jinho Choi, Jaegul Choo, and Steffen Schneider.
Here we develop a new Sparse Autoencoder (SAE) for the CLIP vision transformer, named PatchSAE, to extract interpretable concepts at granular levels (e.g., shape, color, or semantics of an object) and their patch-wise spatial attributions.
[PDF] [code] [poster & video] [website & 🤗 demo]
ICLR 2025 Accepted. Presented in Singapore 🇸🇬
Towards Calibrated Robust Fine-Tuning of Vision-Language Models
Changdae Oh*, Hyesu Lim*, Mijoo Kim, Dongyoon Han, Sangdoo Yun, Jaegul Choo, Alexander G Hauptmann, Zhi-Qi Cheng, Kyungwoo Song. (*: equal contributions)
This work proposes a robust fine-tuning method that improves both OOD accuracy and calibration error in Vision Language Models (VLMs) via fine-tuning with a constrained multimodal contrastive loss that minimizes a shared upper bound of OOD generalization and calibration error. [PDF] [Code]
NeurIPS 2024 Accepted. Presented in Vancouver, Canada 🇨🇦
NeurIPSW 2023 (DistShift) Accepted. Presented in New Orleans, USA 🇺🇲
TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptation
Hyesu Lim, Byeonggeun Kim, Jaegul Choo, and Sungha Choi.
We propose a test-time batch normalization method, which interpolates source and current batch statistics considering each layer's domain-shift sensitivity level, showing robust performance over various realistic evaluation scenarios. [PDF] [website]
ICLR 2023 Accepted. Presented in Kigali, Rwanda 🇷🇼
AVocaDo: Strategy for Adapting Vocabulary to Downstream Domain
Jimin Hong*, Taehee Kim*, Hyesu Lim*, and Jaegul Choo. (*: equal contributions)
We propose to consider a vocabulary of a pre-trained language model as an optimizable parameter, allowing us to update the vocabulary by expanding it with domain-specific vocabulary based on tokenization statistics. [PDF] [Code] [Video&Slide]
EMNLP 2021 Accepted. Presented in Punta Cana, Dominican Republic 🇩🇴
[P02] Adapting machine learning models for domain-shifted data
Hyesu Lim, Byeonggeun Kim, and Sungha Choi
🇺🇲 U.S. Patent No. US20240119360A1, 11 April, 2024. [PDF]
[P01] Suggesting a New and Easier System Function by Detecting User's Action Sequences
Sungrack Yun, Hyoungwoo PARK, Seunghan YANG, Hyesu LIM, Taekyung Kim, Jaewon Choi, Kyu Woong Hwang
🇺🇲 U.S. Patent No. US20240045782A1, 08 Feb, 2024. [PDF]