Skip to main content
Lab directory

KDST (KIST Data Science Team)

KDST (KIST Data Science Team)

Kyung Hee University · Computer Science and Engineering

Information sourcenot verified yet

Ask current students about this lab

Post your question to the Career board — students who have verified their school can answer under blind identities.

Professor

Prof. 김수현

Professor's page · publications

About

김수현 교수 연구실(KDST)은 경희대 컴퓨터공학과 소속으로, 딥러닝과 데이터·지능 전반을 연구한다. 컴퓨터 비전, 대규모 언어 모델(LLM), 디퓨전 모델, 도메인 적응 등을 다루며 파라미터 효율적 미세조정과 추론 효율화 등을 주제로 ECCV·ICLR·NeurIPS 등 최상위 학회에 논문을 발표한다.

Research keywords

Deep learning / AI: model compression & pruningknowledge distillationgenerative & diffusion modelsAI safety

Selected publications

  • Jailbreaking on Text-to-Video Models via Scene Splitting StrategyICLR · 2026
  • Rethinking Entropy in Test-Time Adaptation: The Missing Piece from Energy DualityNeurIPS · 2025
  • Adversarial Purification via Super-Resolution and DiffusionICCV · 2025
  • ELITE: Enhanced Language-Image Toxicity Evaluation for SafetyICML · 2025
  • Difference Inversion: Interpolate and Isolate the Difference with Token Consistency for Image Analogy GenerationCVPR · 2025
  • DEPrune: Depth-wise Separable Convolution Pruning for Maximizing GPU ParallelismNeurIPS · 2024

Other labs in this department

See all labs in this department