Elliot Paquette
8 papers · 2021–2025 · 4 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+1 more ↓ Show less ↑
🌍 Conference Polyglot (4) 🌉 Interdisciplinary Bridge 🗺️ Taxonomy Completionist (13) 🧭 Keyword Pioneer 🐝 Cross-Pollinator (15)
❓
The Questioner
Conferences
NIPS (5)
COLT (1)
ICLR (1)
ICML (1)
Top co-authors
Keywords
stochastic gradient descent
(5)
random matrix theory
(2)
momentum method
(2)
least square
(2)
convergence rate
(2)
adaptive learning rate
(1)
mini-batch optimization
(1)
phase transition
(1)
implicit regularization
(1)
stochastic differential equation
(1)
generalization performance
(1)
nesterov acceleration
(1)
batch size
(1)
data complexity
(1)
convex quadratic
(1)
model parameter
(1)
neural scaling law
(1)
heavy-ball method
(1)
loss curve
(1)
compute-optimal scaling
(1)
Papers
To Clip or not to Clip: the Dynamics of SGD with Gradient Clipping in High-Dimensions
ICLR 2025
Exact risk curves of signSGD in High-Dimensions: quantifying preconditioning and noise-compression effects
ICML 2025
4+3 Phases of Compute-Optimal Neural Scaling Laws
NIPS 2024
The High Line: Exact Risk and Learning Rate Curves of Stochastic Adaptive Learning Rate Algorithms
NIPS 2024
Trajectory of Mini-Batch Momentum: Batch Size Saturation and Convergence in High Dimensions
NIPS 2022
Implicit Regularization or Implicit Conditioning? Exact Risk Trajectories of SGD in High Dimensions
NIPS 2022
Dynamics of Stochastic Momentum Methods on Large-scale, Quadratic Models
NIPS 2021
SGD in the Large: Average-case Analysis, Asymptotics, and Stepsize Criticality
COLT 2021