Prof. Dr. Semih Cayci

Research Interests

Mathematical Foundations of Machine Learning, Deep Learning Theory, Reinforcement Learning, Applied Probability, Optimization Theory

Curriculum Vitae

  • Since 2022: Tenure-Track Assistant Professor for Mathematics of Machine Learning, RWTH Aachen University
  • 2020-2022: NSF TRIPODS Postdoctoral Research Fellow, University of Illinois at Urbana-Champaign
  • 2014-2020: Doctor of Philosophy, Department of Electrical and Computer Engineering, the Ohio State University
  • Teaching

  • Mathematical Foundations of Machine Learning - Winter 2025/26
  • Mathematical Foundations of Deep Learning - Summer 2026, Winter 2023/24
  • Mathematical Foundations of Reinforcement Learning - Winter 2022/23, Summer 2024/25
  • Stochastic Analysis - Winter 2024/25
  • Seminar: Mathematical Learning - Summer 2023
  • News

  • 09.06.2026 - I have received an NVIDIA Academic Research Grant for the project 'GPU-Accelerated Benchmarking of Learning-Based Routing under Heavy Tails'.
  • 30.04.2026 - Geometric Convergence of Gauss–Newton for Neural Networks: Riemannian Geometry and Adaptive Damping was accepted to ICML 2026.
  • 30.04.2026 - I will give a tutorial on policy gradient methods at Reinforcement Learning Summer School in Milano on June 5th, 2026.
  • 11.03.2026 - I will give a talk about RL for POMDPs at ECC Workshop on Learning in Dynamic Games under Partial Information.
  • 22.01.2026 - Two papers got accepted to AISTATS 2026: Non-Asymptotic Generalization and Optimization Bounds for Stochastic Gauss-Newton in Deep Neural Networks [Paper] and Finite-Time Analysis of Gradient Descent for Shallow Transformers [Paper].
  • 29.11.2025 - Optimal Rates of Convergence for Entropy Regularization in Discounted Markov Decision Processes was accepted to Information and Inference: A Journal of the IMA. [Paper]
  • 09.10.2025 - Recurrent Natural Policy Gradient for POMDPs was published in Transactions on Machine Learning Research. [Paper][Slides]
  • 26.07.2025 - Budgeted-Bandits with Controlled Restarts with Applications in Learning and Computing was published in Transactions on Machine Learning Research. [Paper]
  • 23.06.2025 - Convergence of Gradient Descent for Recurrent Neural Networks: A Nonasymptotic Analysis was published in SIAM Journal on Math of Data Science. [Paper]
  • 15.05.2025 - Fisher–Rao Gradient Flows of Linear Programs and State-Action Natural Policy Gradients was published in SIAM Journal on Optimization. [Paper]
  • Contact Options and Office Hours

    Office: Room 202, Pontdriesch 14

    Phone: +49 241 80-97828

    E-mail: cayci@mathc.rwth-aachen.de

    Office Hours:
    by arrangement

    RWTH Aachen University

    Chair of Analysis and its Applications

    Attn: Prof. Dr. Semih Cayci

    Pontdriesch 14

    52062 Aachen

    Germany

    Valid HTML5 · Valid CSS3