Seminar

Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)
Can layerwise linear models simplify complex neural network dynamics and speed up deep learning research?
Jul 4, 2025 2:00 PM — 3:00 PM Room 526, Asan Science Building
ICML ArXiv
Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)
Korea University Physics Prof. Alex Rothkopf's Talk
The talk will be about AI and quantum computing
Jun 30, 2025 1:00 PM — 2:00 PM Room 526, Asan Science Building
Korea University Physics Prof. Alex Rothkopf's Talk
Autoformalization and Automated Theorem Proving with Language Models
Can language models translate natural language math into formal proofs (autoformalization) and more?
Jun 27, 2025 1:30 PM — 2:30 PM Room 526, Asan Science Building
KMS
Autoformalization and Automated Theorem Proving with Language Models
An Exactly Solvable Model for Emergence and Scaling Laws in the Multitask Sparse Parity Problem
Can an analytically solvable model explain emergent behaviors and scaling laws in multitask learning, showing how new skills appear as training progresses?
Jun 17, 2025 1:00 PM — 2:00 PM Room 526, Asan Science Building
NeurIPS Poster ArXiv
An Exactly Solvable Model for Emergence and Scaling Laws in the Multitask Sparse Parity Problem
Formalizing Mathematics: Why and How
Formalization in mathematics – translating math into computer-readable language – is growing into successful projects!
May 26, 2025 4:00 PM — 5:00 PM Room 526, Asan Science Building
KMS
Formalizing Mathematics: Why and How