Felix Dangel

Assistant Professor@Concordia University·加拿大·OpenReview

研究方向

Influence functions · Scientific Machine Learning · Automatic Differentiation · Deep Learning optimization · Second-order optimization · Tensor networks

Felix Dangel

Understanding and improving Shampoo and SOAP via Kullback-Leibler Minimization

Dataless Weight Disentanglement in Task Arithmetic via Kronecker-Factored Approximate Curvature

Improving Energy Natural Gradient Descent through Woodbury, Momentum, and Randomization

What Does It Mean to Be a Transformer? Insights from a Theoretical Hessian Analysis

Collapsing Taylor Mode Automatic Differentiation

Fishers for Free? Approximating the Fisher Information Matrix by Recycling the Squared Gradient Accumulator

Hide & Seek: Transformer Symmetries Obscure Sharpness & Riemannian Geometry Finds It

Hide & Seek: Transformer Symmetries Obscure Sharpness & Riemmanian Geometry Finds It

Fast Fractional Natural Gradient Descent using Learnable Spectral Factorizations