From: NeurIPS 2024 Program Committee notifications@neurips.cc To: Alex Vukovic alexvuk@ucsb.edu Date: September 3, 2025 Subject: NeurIPS 2024 Paper Acceptance Notification
Dear Alex,
We are pleased to inform you that your submission "Thermodynamic Bounds on Learned Representations: An Information-Geometric Approach" has been accepted for presentation at NeurIPS 2024...
Alex reads the email twice. Then a third time, standing in the grad student lounge, his coffee going cold in his hand.
Accepted.
Not just accepted - accepted with strong reviewer scores. One reviewer called it "a novel contribution bridging statistical mechanics and representation learning." Another said "the entropy reduction framework is elegant and well-motivated." The third suggested it could "open new directions in theoretical ML."
He sits down. Hard. The plastic chair creaks under him.
All those nights. All those weekends when he could have been at parties, on dates, doing literally anything else. The spring semester grinding through measure theory and differential geometry while his classmates coasted. The summer in the library, sweating through his shirt because the AC was broken, Marchetti's voice in his ear from recorded lectures, pushing him to be rigorous, to be precise, to not cut corners.
It paid off.