3 papers
cs.CL2026
XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity
Dasol Choi, Eugenia Kim, Jaewon Noh +14
Current LLM safety benchmarks are predominantly English-centric and often rely on translation, failing to capture country-specific harms. Moreover, they rarely evaluate a model's a…
cs.CV2024
Adversarial Feature Alignment: Balancing Robustness and Accuracy in Deep Learning via Adversarial Training
Leo Hyun Park, Jaeuk Kim, Myung Gyo Oh +2
Deep learning models continue to advance in accuracy, yet they remain vulnerable to adversarial attacks, which often lead to the misclassification of adversarial examples. Adversar…
cs.CL2024
Amplifying Training Data Exposure through Fine-Tuning with Pseudo-Labeled Memberships
Myung Gyo Oh, Hong Eun Ahn, Leo Hyun Park +1
Neural language models (LMs) are vulnerable to training data extraction attacks due to data memorization. This paper introduces a novel attack scenario wherein an attacker adversar…