Tom Jacobs
Scholar

Tom Jacobs

Google Scholar ID: Yi62zDEAAAAJ
PhD student, CISPA Helmholtz Center for Information Security
Training DynamicsDeep LearningOptimization
Citations & Impact
All-time
Citations
3
 
H-index
1
 
i10-index
0
 
Publications
6
 
Co-authors
1
list available
Resume (English only)
Academic Achievements
  • First-author paper 'Sign-In to the Lottery: Reparameterizing Sparse Training From Scratch' accepted at NeurIPS 2025 (San Diego); two co-authored papers 'The Graphon Limit Hypothesis: Understanding Neural Network Pruning via Infinite Width Analysis' (Spotlight) and 'Pay Attention to Small Weights' also accepted
  • First-author paper 'Mirror, Mirror of the Flow: How Does Regularization Shape Implicit Bias?' accepted at ICML 2025 (Vancouver)
  • First-author paper 'Mask in the Mirror: Implicit Sparsification' accepted at ICLR 2025 (Singapore)
  • Passed PhD qualifying exam (QE) on January 6, 2025; committee members: Dr. Sebastian Stich, Dr. Antonio Krüger, and advisor Dr. Rebekka Burkholz
  • Released first-author preprint 'HAM: A Hyperbolic Step to Regulate Implicit Bias' on June 4, 2025