Statistical Inference Lab
Statistical Inference Lab
KAIST · Graduate School of AI
This lab is recruiting graduate students
연구실에 관심 있는 학생은 홈페이지의 application page(Notion 기반 지원 포털)를 참고하도록 안내함. KAIST 김재철AI대학원 입학 필요.
Check the professor and homepage links below for how to reach out.
Professor
Prof. 윤세영
Professor's page · publicationsAbout
윤세영 교수가 이끄는 OSI(Optimization and Statistical Inference) 연구실은 KAIST 김재철AI대학원 서울 홍릉캠퍼스에 위치하며, 이론적 분석을 바탕으로 학습 알고리즘의 효율성과 일반화 성능을 이해·개선하는 연구를 수행한다. 강화학습 이론부터 대규모 언어모델 최적화·증류, 생성 AI, 멀티모달 머신러닝까지 폭넓은 주제를 다룬다.
Research keywords
Recent papers
- Beyond Second-Order Information: Paid Logistic Regression with Noisy FeaturesNewPreprintarXiv (Cornell University) · Oct 9
- Conversation Is a Two-Body Problem: Dyadic Evaluation of Full-Duplex Dialogue ModelsPreprintarXiv (Cornell University) · Oct 6
- Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language ModelsPreprintarXiv (Cornell University) · Oct 2
From OpenAlex · papers whose author affiliation matches this university
Selected publications
- Mixture-of-Recursions: Learning Dynamic Recursive Depths for Adaptive Token-Level ComputationNeurIPS · 2025
- KLASS: KL-Guided Fast Inference in Masked Diffusion ModelsNeurIPS (Spotlight) · 2025
- Flex-Judge: Text-Only Reasoning Unleashes Zero-Shot Multimodal EvaluatorsNeurIPS · 2025
- DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMsICML (Oral) · 2025
- MoHAVE: Mixture of Hierarchical Audio-Visual Experts for Robust Speech RecognitionICML · 2025
- DistiLLM: Towards Streamlined Distillation for Large Language ModelsICML · 2024
- Hard Prompts Made Interpretable: Sparse Entropy Regularization for Prompt Tuning with RLACL (Oral) · 2024
Labs doing related research
Labs that recently published on topics this lab's recent papers share
Kim Junmo Lab
KAIST · Electrical Engineering
Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-TrainingNewPreprintarXiv (Cornell University) · Sep 9Shared topic: Reinforcement Learning in Robotics
Yoo Changdong Lab
KAIST · Electrical Engineering
Progress-Guided Preference Generation for Aligning Generative Control PoliciesNewPreprintarXiv (Cornell University) · Oct 19Shared topic: Reinforcement Learning in Robotics
Artificial Intelligence and Information Systems Lab (AI-ISL)
Yonsei University · Computer Science
Low-Confidence Remasking Traps Flexibility: Realizing Arbitrary-Order Potential for Diverse Rollouts in Diffusion LLMsPreprintarXiv (Cornell University) · Sep 28Shared topic: Topic Modeling
Other labs in this department
- AI and Machine Learning LabRecruiting
Kee-Eung Kim
Artificial Intelligence, Machine Learning, Reinforcement Learning
- Robotics and ML LabRecruiting
Beomjoon Kim
Robotics, machine learning, planning
- Sehoon KimRecruiting
AI systems, efficient AI, large language models
- Computer Vision LabRecruiting
Seungryong Kim
Computer vision, machine learning
- Visual Intelligence LabRecruiting
Hyunjung Shim
Computer vision, machine learning
- Probabilistic AI LabRecruiting
Eunho Yang
Machine learning, artificial intelligence