Kwangjun Ahn
21 papers · 2018–2025 · 6 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+9 more ↓ Show less ↑
π Cross-Pollinator (8) π Academic Marathon (7) π§ Keyword Pioneer π Conference Polyglot (6) π Renaissance Researcher (5)
π
Conference Polyglot
(6)
π
Academic Marathon
(7)
π§
Keyword Pioneer
π
Triple Crown
π₯
Unstoppable
(6)
β‘
Prolific Year
(5)
π
Century Club
(21)
β
The Questioner
ποΈ
Keyword Collector
(74)
Conferences
NIPS (9)
ICML (5)
ICLR (3)
COLT (2)
JMLR (1)
L4DC (1)
Top co-authors
Keywords
gradient descent
(3)
langevin dynamics
(2)
mirror descent
(2)
convergence rate
(2)
stochastic gradient descent
(2)
non-convex optimization
(2)
maximum-margin solution
(2)
log-concave distribution
(2)
nonconvex optimization
(2)
markov chain monte carlo
(2)
convex optimization
(2)
global convergence
(1)
in-context learning
(1)
convergence analysis
(1)
optimal transport
(1)
finite-sum optimization
(1)
collaborative filtering
(1)
logistic regression
(1)
neural network optimization
(1)
model predictive control
(1)
Papers
The Belief State Transformer
ICLR 2025
General framework for online-to-nonconvex conversion: Schedule-free SGD is also effective for nonconvex optimization
ICML 2025
Does SGD really happen in tiny subspaces?
ICLR 2025
How to Escape Sharp Minima with Random Perturbations
ICML 2024
Understanding Adam Optimizer via Online Learning of Updates: Adam is FTRL in Disguise
ICML 2024
Adam with model exponential moving average is effective for nonconvex optimization
NIPS 2024
Linear attention is (maybe) all you need (to understand Transformer optimization)
ICLR 2024
The Crucial Role of Normalization in Sharpness-Aware Minimization
NIPS 2023
A Unified Approach to Controlling Implicit Regularization via Mirror Descent
JMLR 2023
Model Predictive Control via On-Policy Imitation Learning
L4DC 2023
Learning threshold neurons via edge of stability
NIPS 2023
Transformers learn to implement preconditioned gradient descent for in-context learning
NIPS 2023
Understanding the unstable convergence of gradient descent
ICML 2022
Reproducibility in Optimization: Theoretical Framework and Limits
NIPS 2022
Mirror Descent Maximizes Generalized Margin and Can Be Implemented Efficiently
NIPS 2022
Agnostic Learnability of Halfspaces via Logistic Loss
ICML 2022
Optimal dimension dependence of the Metropolis-Adjusted Langevin Algorithm
COLT 2021
Efficient constrained sampling via the mirror-Langevin algorithm
NIPS 2021
From Nesterovβs Estimate Sequence to Riemannian Acceleration
COLT 2020
SGD with shuffling: optimal rates without component convexity and large epoch requirements
NIPS 2020
Binary Rating Estimation with Graph Side Information
NIPS 2018