Oğuz Kaan Yüksel
  • Research
  • Publications

Oğuz Kaan Yüksel

I am a fourth-year Ph.D candidate at the Theory of Machine Learning Lab, EPFL, advised by Nicolas Flammarion. I focus on understanding AI from first principles — using mathematical modeling and rigorous theory to explain how and why modern learning systems work. My work spans generalization, optimization, and the structure of data, with a particular interest in language models and autoregressive processes.

Research · CV

Background. M.Sc. in Data Science (minor in Mathematics) and Ph.D. at EPFL. B.Sc. in Computer Engineering & Mathematics at Boğaziçi University.

Selected Publications

Citation copied to clipboard
Incremental Learning of Sparse Attention Patterns in Transformers
Incremental Learning of Sparse Attention Patterns in Transformers
Oğuz Kaan Yüksel, Rodrigo Alvarez Lucendo, Nicolas Flammarion
In ICML International Conference on Machine Learning , 2026
Paper arXiv Slides
ICML
Induction Heads Interpolate N-Grams
Francesco D'Angelo*Equal contribution, Oğuz Kaan Yüksel*Equal contribution, Swathi Shree Narashiman, Nicolas Flammarion
In ICML International Conference on Machine Learning , 2026
Paper
No matching items

All publications →

News

Apr 30, 2026 paper

Two papers [Incremental Learning of Sparse Attention Patterns in Transformers](/publications/yuksel2026incremental) and [Induction Heads Interpolate N-Grams](/publications/dangelo2026induction) are accepted to ICML, 2026.

Jul 12, 2025 talk

Presented a talk on [Incremental Learning of Sparse Attention Patterns in Transformers](/slides/prigm-2025/) at PriGM, EurIPS 2025.

Jul 12, 2025 poster

Presented two posters [Incremental Learning of Sparse Attention Patterns in Transformers](/publications/yuksel2025incremental) and [Generalization Bounds for Autoregressive Processes and In-Context Learning](/publications/yuksel2025generalization) at PriGM, EurIPS 2025.

May 3, 2025 poster

Presented [On the Sample Complexity of Next-Token Prediction](/publications/yuksel2025sample) at AISTATS 2025.

Apr 24, 2025 poster

Presented [Long-Context Linear System Identification](/publications/yuksel2025long) at ICLR 2025.

May 7, 2024 poster

Presented [First-order ANIL provably learns representations despite overparametrization](/publications/yuksel2024first) at ICLR 2024.

Nov 16, 2023 poster

Presented [First-order ANIL provably learns representations despite overparametrization](/publications/yuksel2024first) at NeurIPS 2023.

No matching items