OSI LAB

Welcome

Optimization and Statistical Inference Lab

Welcome to the Optimization and Statistical Inference Lab at KAIST AI. We tackle a wide range of problems, from theoretical analysis of traditional machine learning, to the latest research topics in deep learning. Please refer to publications for details.

If you are interested in joining us, please refer here.

Group Photo Spring 2024

7 papers accepted to ICLR 2025!

Automated Filtering of Human Feedback Data for Aligning Text-to-Image Diffusion Models
FlickerFusion: Intra-trajectory Domain Generalizing Multi-agent Reinforcement Learning
Multi-Task Corrupted Prediction for Learning Robust Audio-Visual Speech Representation
MA2E: Addressing Partial Observability in Multi-Agent Reinforcement Learning with Masked Auto-Encoder
Relaxed Recursive Transformers: Effective Parameter Sharing with Layer-wise LoRA
SeRA: Self-Reviewing and Alignment of LLMs using Implicit Reward Margins
Beyond Correlation: The Impact of Human Uncertainty in Measuring the Effectiveness of Automatic Evaluation and LLM-as-a-Judge

1 paper accepted to AISTATS 2025!

Bayesian Principles Improve Prompt Learning in Vision-Language Models

2 papers accepted to NAACL 2025! (1 main, 1 findings)

Main:

C^2: Scalable Auto-Feedback for LLM-based Chart Generation

Findings:

MAQA: Evaluating Uncertainty Quantification in LLMs Regarding Data Uncertainty

1 paper at QIFK 2024 selected as one of the *Winners*

DistiLLM: Towards Streamlined Distillation for Large Language Models