[Oct 25, 2024]
1 paper accepted to TMLR!
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.
Latest News
See More
[Sep 26, 2024]
8 papers accepted to NeurIPS 2024 (6 main, 2 workshop)
Main:
- Conditional Synthesis of 3D Molecules with Time Correction Sampler
- An Adaptive Approach for Infinitely Many-armed Bandits under Generalized Rotting Constraints
- A Unified Confidence Sequence for Generalized Linear Models, with Applications to Bandits
- Preference Alignment with Flow Matching
- Block Transformer: Global-to-Local Language Modeling for Fast Inference
- Exploring & Improving Multi-token Prediction (Block Draft) in Language Modeling
Workshop:
- A Unified Framework for Speculative Decoding with Multiple Drafters as a Bandit (ENSLP Workshop)
- FlickerFusion: Intra-trajectory Domain Generalizing Multi-agent Reinforcement Learning (OWA Workshop)
[Sep 20, 2024]
5 papers accepted to EMNLP 2024 (4 main, 1 finding)
Main:
- Towards Fast Multilingual LLM Inference: Speculative Decoding and Specialized Drafters
- Stable Language Model Pre-training by Reducing Embedding Variability
- DocKD: Knowledge Distillation from LLMs for Open-World Document Understanding Models
- Towards Difficulty-Agnostic Efficient Transfer Learning for Vision-Language Models
Findings: