Kwangjun Ahn

Engineering · Physical Sciences

12h-index976citations65works0.82yr avg
OpenAlex↗ORCID↗Google Scholar↗

Accepting Students?

No reports yet. Know if this professor is taking students?

Research Topics

Plasmonic and Surface Plasmon Research(15), Stochastic Gradient Optimization Techniques(13), Sparse and Compressive Sensing Techniques(10), Photonic and Optical Devices(7), Photonic Crystals and Applications(6)

Publications65 total

Next-Latent Prediction Transformers Learn Compact World Models
ArXiv.org·2025Open Access
Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
arXiv (Cornell University)·2025Open Access
Dion: Distributed Orthonormalized Updates
ArXiv.org·2025Open Access
Efficient Joint Prediction of Multiple Future Tokens
ArXiv.org·2025Open Access
On the Sample Complexity of Imitation Learning for Smoothed Model Predictive Control
2024· 1 cited
General framework for online-to-nonconvex conversion: Schedule-free SGD is also effective for nonconvex optimization
arXiv (Cornell University)·2024Open Access
The Belief State Transformer
arXiv (Cornell University)·2024Open Access
Improved Sample Complexity of Imitation Learning for Barrier Model Predictive Control
arXiv (Cornell University)·2024Open Access
Adam with model exponential moving average is effective for nonconvex optimization
arXiv (Cornell University)·2024· 2 citedOpen Access
Does SGD really happen in tiny subspaces?
arXiv (Cornell University)·2024Open Access
Understanding Adam Optimizer via Online Learning of Updates: Adam is FTRL in Disguise
arXiv (Cornell University)·2024· 2 citedOpen Access
Adam with model exponential moving average is effective for nonconvex optimization
2024· 5 cited
Linear attention is (maybe) all you need (to understand transformer optimization)
arXiv (Cornell University)·2023· 1 citedOpen Access
A Unified Approach to Controlling Implicit Regularization via Mirror Descent
arXiv (Cornell University)·2023Open Access
On the Sample Complexity of Imitation Learning for Smoothed Model Predictive Control
arXiv (Cornell University)·2023Open Access
Transformers learn to implement preconditioned gradient descent for in-context learning
arXiv (Cornell University)·2023· 6 citedOpen Access
How to escape sharp minima with random perturbations
arXiv (Cornell University)·2023Open Access
The Crucial Role of Normalization in Sharpness-Aware Minimization
arXiv (Cornell University)·2023Open Access
Learning threshold neurons via the "edge of stability"
arXiv (Cornell University)·2022Open Access
One-Pass Learning via Bridging Orthogonal Gradient Descent and Recursive Least-Squares
2022 IEEE 61st Conference on Decision and Control (CDC)·2022· 9 cited
page 1 of 4Next →

Frequent Co-authors

Suvrit Sra(8), Minah Seo(6), Namkyoo Park(5), Kangwook Lee(5), Changho Suh(5), Jisoo Kyoung(4), Dai‐Sik Kim(4), D. S. Kim(4), Chulhee Yun(4), Ali Jadbabaie(4), Sukmo Koo(3), Kinam Jung(3), Kyungyeon Ha(3), Jong‐Kwon Lee(3), H. R. Park(3), A. J. L. Adam(3), P. C. M. Planken(3), Navid Azizan(3), Sinho Chewi(3), Zakaria Mhammedi(3)