Publications

2026

1. Geometric Convergence of Gauss–Newton for Neural Networks: Riemannian Geometry and Adaptive Damping

   Semih Cayci

   In Proceedings of the 43rd International Conference on Machine Learning, 2026

   HTML Bib

   @inproceedings{cayci2026geometric,
     author = {Cayci, Semih},
     title = {{Geometric Convergence of Gauss--Newton for Neural Networks: Riemannian Geometry and Adaptive Damping}},
     booktitle = {Proceedings of the 43rd International Conference on Machine Learning},
     series = {Proceedings of Machine Learning Research},
     volume = {306},
     pages = {12189--12214},
     year = {2026},
     publisher = {PMLR},
     url = {https://proceedings.mlr.press/v306/cayci26a.html},
   }

2. Non-Asymptotic Generalization and Optimization Bounds for Stochastic Gauss-Newton in Deep Neural Networks

   Semih Cayci

   In Proceedings of the 29th International Conference on Artificial Intelligence and Statistics, 2026

   HTML Bib

   @inproceedings{cayci2026stochasticgaussnewton,
     author = {Cayci, Semih},
     title = {{Non-Asymptotic Generalization and Optimization Bounds for Stochastic Gauss-Newton in Deep Neural Networks}},
     booktitle = {Proceedings of the 29th International Conference on Artificial Intelligence and Statistics},
     series = {Proceedings of Machine Learning Research},
     volume = {300},
     pages = {1549--1557},
     year = {2026},
     publisher = {PMLR},
     url = {https://proceedings.mlr.press/v300/cayci26a.html},
   }

3. Optimal Rates of Convergence for Entropy Regularization in Discounted Markov Decision Processes

   Johannes Müller and Semih Cayci

   Information and Inference: A Journal of the IMA, 2026

   DOI Bib

   @article{muller2026optimal,
     author = {M{\"u}ller, Johannes and Cayci, Semih},
     title = {{Optimal Rates of Convergence for Entropy Regularization in Discounted Markov Decision Processes}},
     journal = {Information and Inference: A Journal of the IMA},
     volume = {15},
     number = {1},
     pages = {iaaf034},
     year = {2026},
     doi = {10.1093/imaiai/iaaf034},
     url = {https://doi.org/10.1093/imaiai/iaaf034},
   }

4. Convergence of Stochastic Gradient Langevin Dynamics in the Lazy Training Regime

   Noah Oberweis and Semih Cayci

   Transactions on Machine Learning Research, 2026

   HTML PDF Bib

   @article{oberweis2026langevin,
     author = {Oberweis, Noah and Cayci, Semih},
     title = {{Convergence of Stochastic Gradient Langevin Dynamics in the Lazy Training Regime}},
     journal = {Transactions on Machine Learning Research},
     year = {2026},
     url = {https://openreview.net/forum?id=3s037DKmLo},
   }

5. Finite-Time Analysis of Gradient Descent for Shallow Transformers

   Enes Arda, Semih Cayci, and Atilla Eryilmaz

   In Proceedings of the 29th International Conference on Artificial Intelligence and Statistics, 2026

   HTML Bib

   @inproceedings{arda2026transformers,
     author = {Arda, Enes and Cayci, Semih and Eryilmaz, Atilla},
     title = {{Finite-Time Analysis of Gradient Descent for Shallow Transformers}},
     booktitle = {Proceedings of the 29th International Conference on Artificial Intelligence and Statistics},
     series = {Proceedings of Machine Learning Research},
     volume = {300},
     pages = {3988--3996},
     year = {2026},
     publisher = {PMLR},
     url = {https://proceedings.mlr.press/v300/arda26a.html},
   }

6. Machine-learned R13 moment closures for shock-dominated rarefied gas flows

   Hang Song, Satyvir Singh, Semih Cayci, and Manuel Torrilhon

   Computers & Fluids, 2026

   DOI Bib

   @article{song2026r13,
     author = {Song, Hang and Singh, Satyvir and Cayci, Semih and Torrilhon, Manuel},
     title = {{Machine-learned R13 moment closures for shock-dominated rarefied gas flows}},
     journal = {Computers \& Fluids},
     volume = {319},
     pages = {107260},
     year = {2026},
     doi = {10.1016/j.compfluid.2026.107260},
     url = {https://doi.org/10.1016/j.compfluid.2026.107260},
   }

2025

1. Convergence of Gradient Descent for Recurrent Neural Networks: A Nonasymptotic Analysis

   Semih Cayci and Atilla Eryilmaz

   SIAM Journal on Mathematics of Data Science, 2025

   DOI Bib

   @article{cayci2025rnngradient,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{Convergence of Gradient Descent for Recurrent Neural Networks: A Nonasymptotic Analysis}},
     journal = {SIAM Journal on Mathematics of Data Science},
     volume = {7},
     number = {2},
     pages = {826--854},
     year = {2025},
     doi = {10.1137/24M1642500},
     url = {https://doi.org/10.1137/24M1642500},
   }

2. Recurrent Natural Policy Gradient for POMDPs

   Semih Cayci and Atilla Eryilmaz

   Transactions on Machine Learning Research, 2025

   HTML PDF Code Bib

   @article{cayci2025recurrent,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{Recurrent Natural Policy Gradient for POMDPs}},
     journal = {Transactions on Machine Learning Research},
     year = {2025},
     url = {https://openreview.net/forum?id=6G01e0vgIf},
   }

3. Fisher–Rao Gradient Flows of Linear Programs and State-Action Natural Policy Gradients

   Johannes Müller, Semih Cayci, and Guido Montúfar

   SIAM Journal on Optimization, 2025

   DOI Bib

   @article{muller2025fisher,
     author = {M{\"u}ller, Johannes and Cayci, Semih and Mont{\'u}far, Guido},
     title = {{Fisher--Rao Gradient Flows of Linear Programs and State-Action Natural Policy Gradients}},
     journal = {SIAM Journal on Optimization},
     volume = {35},
     number = {2},
     pages = {1060--1088},
     year = {2025},
     doi = {10.1137/24M1653422},
     url = {https://doi.org/10.1137/24M1653422},
   }

4. Budgeted-Bandits with Controlled Restarts with Applications in Learning and Computing

   Semih Cayci, Yilin Zheng, and Atilla Eryilmaz

   Transactions on Machine Learning Research, 2025

   HTML PDF Bib

   @article{cayci2025budgeted,
     author = {Cayci, Semih and Zheng, Yilin and Eryilmaz, Atilla},
     title = {{Budgeted-Bandits with Controlled Restarts with Applications in Learning and Computing}},
     journal = {Transactions on Machine Learning Research},
     year = {2025},
     url = {https://openreview.net/forum?id=lvb5qDAa4B},
   }

5. A Variational Inequality Approach to Independent Learning in Static Mean-Field Games

   Batuhan Yardim, Semih Cayci, and Niao He

   ACM/IMS Journal of Data Science, 2025

   DOI Bib

   @article{yardim2025variational,
     author = {Yardim, Batuhan and Cayci, Semih and He, Niao},
     title = {{A Variational Inequality Approach to Independent Learning in Static Mean-Field Games}},
     journal = {ACM/IMS Journal of Data Science},
     volume = {2},
     number = {2},
     pages = {7:1--7:43},
     year = {2025},
     doi = {10.1145/3731553},
     url = {https://doi.org/10.1145/3731553},
   }

2024

1. Convergence of Entropy-Regularized Natural Policy Gradient with Linear Function Approximation

   Semih Cayci, Niao He, and R. Srikant

   SIAM Journal on Optimization, 2024

   DOI Bib

   @article{cayci2024convergence,
     author = {Cayci, Semih and He, Niao and Srikant, R.},
     title = {{Convergence of Entropy-Regularized Natural Policy Gradient with Linear Function Approximation}},
     journal = {SIAM Journal on Optimization},
     volume = {34},
     number = {3},
     pages = {2729--2755},
     year = {2024},
     doi = {10.1137/22M1540156},
     url = {https://doi.org/10.1137/22M1540156},
   }

2. Finite-Time Analysis of Natural Actor-Critic for POMDPs

   Semih Cayci, Niao He, and R. Srikant

   SIAM Journal on Mathematics of Data Science, 2024

   DOI Bib

   @article{cayci2024finite,
     author = {Cayci, Semih and He, Niao and Srikant, R.},
     title = {{Finite-Time Analysis of Natural Actor-Critic for POMDPs}},
     journal = {SIAM Journal on Mathematics of Data Science},
     volume = {6},
     number = {4},
     pages = {869--896},
     year = {2024},
     doi = {10.1137/23M1587683},
     url = {https://doi.org/10.1137/23M1587683},
   }

3. Finite-Time Analysis of Entropy-Regularized Neural Natural Actor-Critic Algorithm

   Semih Cayci, Niao He, and R. Srikant

   Transactions on Machine Learning Research, 2024

   HTML PDF Bib

   @article{cayci2024finitetime,
     author = {Cayci, Semih and He, Niao and Srikant, R.},
     title = {{Finite-Time Analysis of Entropy-Regularized Neural Natural Actor-Critic Algorithm}},
     journal = {Transactions on Machine Learning Research},
     year = {2024},
     url = {https://openreview.net/forum?id=BkEqk7pS1I},
   }

4. Fast Online Learning of Vulnerabilities for Networks With Propagating Failures

   Yilin Zheng, Semih Cayci, and Atilla Eryilmaz

   IEEE/ACM Transactions on Networking, 2024

   DOI Bib

   @article{zheng2024vulnerabilities,
     author = {Zheng, Yilin and Cayci, Semih and Eryilmaz, Atilla},
     title = {{Fast Online Learning of Vulnerabilities for Networks With Propagating Failures}},
     journal = {IEEE/ACM Transactions on Networking},
     volume = {32},
     number = {5},
     pages = {4025--4039},
     year = {2024},
     doi = {10.1109/TNET.2024.3405798},
     url = {https://doi.org/10.1109/TNET.2024.3405798},
   }

2023

1. Sample Complexity and Overparameterization Bounds for Temporal-Difference Learning With Neural Network Approximation

   Semih Cayci, Siddhartha Satpathi, Niao He, and R. Srikant

   IEEE Transactions on Automatic Control, 2023

   DOI Bib

   @article{cayci2023sample,
     author = {Cayci, Semih and Satpathi, Siddhartha and He, Niao and Srikant, R.},
     title = {{Sample Complexity and Overparameterization Bounds for Temporal-Difference Learning With Neural Network Approximation}},
     journal = {IEEE Transactions on Automatic Control},
     volume = {68},
     number = {5},
     pages = {2891--2905},
     year = {2023},
     doi = {10.1109/TAC.2023.3234234},
     url = {https://doi.org/10.1109/TAC.2023.3234234},
   }

2. Provably Robust Temporal Difference Learning for Heavy-Tailed Rewards

   Semih Cayci and Atilla Eryilmaz

   In Advances in Neural Information Processing Systems, 2023

   DOI HTML Bib

   @inproceedings{cayci2023provably,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{Provably Robust Temporal Difference Learning for Heavy-Tailed Rewards}},
     booktitle = {Advances in Neural Information Processing Systems},
     volume = {36},
     pages = {39915--39933},
     year = {2023},
     publisher = {Curran Associates, Inc.},
     doi = {10.52202/075280-1117},
     url = {https://papers.neurips.cc/paper_files/paper/2023/hash/516ca2e9e7bffbb4027a25d9f8838bc9-Abstract-Conference.html},
   }

3. Policy Mirror Ascent for Efficient and Independent Learning in Mean Field Games

   Batuhan Yardim, Semih Cayci, Matthieu Geist, and Niao He

   In Proceedings of the 40th International Conference on Machine Learning, 2023

   HTML Bib

   @inproceedings{yardim2023policy,
     author = {Yardim, Batuhan and Cayci, Semih and Geist, Matthieu and He, Niao},
     title = {{Policy Mirror Ascent for Efficient and Independent Learning in Mean Field Games}},
     booktitle = {Proceedings of the 40th International Conference on Machine Learning},
     series = {Proceedings of Machine Learning Research},
     volume = {202},
     pages = {39722--39754},
     year = {2023},
     publisher = {PMLR},
     url = {https://proceedings.mlr.press/v202/yardim23a.html},
   }

2022

1. A Lyapunov-Based Methodology for Constrained Optimization with Bandit Feedback

   Semih Cayci, Yilin Zheng, and Atilla Eryilmaz

   In Proceedings of the AAAI Conference on Artificial Intelligence, 2022

   DOI HTML Bib

   @inproceedings{cayci2022lyapunov,
     author = {Cayci, Semih and Zheng, Yilin and Eryilmaz, Atilla},
     title = {{A Lyapunov-Based Methodology for Constrained Optimization with Bandit Feedback}},
     booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
     volume = {36},
     number = {4},
     pages = {3716--3723},
     year = {2022},
     doi = {10.1609/aaai.v36i4.20285},
     url = {https://ojs.aaai.org/index.php/AAAI/article/view/20285},
   }

2020

1. Budget-Constrained Bandits over General Cost and Reward Distributions

   Semih Cayci, Atilla Eryilmaz, and R. Srikant

   In Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, 2020

   HTML Bib

   @inproceedings{cayci2020budget,
     author = {Cayci, Semih and Eryilmaz, Atilla and Srikant, R.},
     title = {{Budget-Constrained Bandits over General Cost and Reward Distributions}},
     booktitle = {Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics},
     series = {Proceedings of Machine Learning Research},
     volume = {108},
     pages = {4388--4398},
     year = {2020},
     publisher = {PMLR},
     url = {https://proceedings.mlr.press/v108/cayci20a.html},
   }

2. Group-Fair Online Allocation in Continuous Time

   Semih Cayci, Swati Gupta, and Atilla Eryilmaz

   In Advances in Neural Information Processing Systems, 2020

   HTML Bib

   @inproceedings{cayci2020group,
     author = {Cayci, Semih and Gupta, Swati and Eryilmaz, Atilla},
     title = {{Group-Fair Online Allocation in Continuous Time}},
     booktitle = {Advances in Neural Information Processing Systems},
     volume = {33},
     pages = {13750--13761},
     year = {2020},
     publisher = {Curran Associates, Inc.},
     url = {https://proceedings.neurips.cc/paper_files/paper/2020/hash/9ec0cfdc84044494e10582436e013e64-Abstract.html},
   }

2019

1. Optimal Learning for Dynamic Coding in Deadline-Constrained Multi-Channel Networks

   Semih Cayci and Atilla Eryilmaz

   IEEE/ACM Transactions on Networking, 2019

   DOI Bib

   @article{cayci2019dynamic,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{Optimal Learning for Dynamic Coding in Deadline-Constrained Multi-Channel Networks}},
     journal = {IEEE/ACM Transactions on Networking},
     volume = {27},
     number = {3},
     pages = {1043--1054},
     year = {2019},
     doi = {10.1109/TNET.2019.2913666},
     url = {https://doi.org/10.1109/TNET.2019.2913666},
   }

2. Learning to Control Renewal Processes with Bandit Feedback

   Semih Cayci, Atilla Eryilmaz, and R. Srikant

   Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2019

   ACM SIGMETRICS 2019

   DOI Bib

   @article{cayci2019learning,
     author = {Cayci, Semih and Eryilmaz, Atilla and Srikant, R.},
     title = {{Learning to Control Renewal Processes with Bandit Feedback}},
     journal = {Proceedings of the ACM on Measurement and Analysis of Computing Systems},
     volume = {3},
     number = {2},
     pages = {43:1--43:32},
     year = {2019},
     publisher = {Association for Computing Machinery},
     note = {ACM SIGMETRICS 2019},
     doi = {10.1145/3326158},
     url = {https://doi.org/10.1145/3326158},
   }

3. Nonbinary Polar Coding for Multilevel Modulation

   Semih Cayci, Toshiaki Koike-Akino, and Ye Wang

   In Optical Fiber Communication Conference, 2019

   DOI Bib

   @inproceedings{cayci2019nonbinary,
     author = {Cayci, Semih and Koike-Akino, Toshiaki and Wang, Ye},
     title = {{Nonbinary Polar Coding for Multilevel Modulation}},
     booktitle = {Optical Fiber Communication Conference},
     pages = {W3H.4},
     year = {2019},
     publisher = {Optical Society of America},
     doi = {10.1364/OFC.2019.W3H.4},
     url = {https://doi.org/10.1364/OFC.2019.W3H.4},
   }

4. Hardware-Efficient Quantized Polar Decoding with Optimized Lookup Table

   Toshiaki Koike-Akino, Ye Wang, Semih Cayci, David S. Millar, Keisuke Kojima, and Kieran Parsons

   In 2019 24th OptoElectronics and Communications Conference (OECC) and 2019 International Conference on Photonics in Switching and Computing (PSC), 2019

   DOI Bib

   @inproceedings{koikeakino2019hardware,
     author = {Koike-Akino, Toshiaki and Wang, Ye and Cayci, Semih and Millar, David S. and Kojima, Keisuke and Parsons, Kieran},
     title = {{Hardware-Efficient Quantized Polar Decoding with Optimized Lookup Table}},
     booktitle = {2019 24th OptoElectronics and Communications Conference (OECC) and 2019 International Conference on Photonics in Switching and Computing (PSC)},
     pages = {1--3},
     year = {2019},
     doi = {10.23919/PS.2019.8817861},
     url = {https://doi.org/10.23919/PS.2019.8817861},
   }

2017

1. Learning for Serving Deadline-Constrained Traffic in Multi-Channel Wireless Networks

   Semih Cayci and Atilla Eryilmaz

   In 2017 15th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt), 2017

   DOI Bib

   @inproceedings{cayci2017deadlines,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{Learning for Serving Deadline-Constrained Traffic in Multi-Channel Wireless Networks}},
     booktitle = {2017 15th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt)},
     pages = {1--8},
     year = {2017},
     doi = {10.23919/WIOPT.2017.7959911},
     url = {https://doi.org/10.23919/WIOPT.2017.7959911},
   }

2016

1. On the Multi-Channel Capacity Gains of Millimeter-Wave Communication

   Semih Cayci and Atilla Eryilmaz

   In 2016 IEEE Global Communications Conference (GLOBECOM), 2016

   DOI Bib

   @inproceedings{cayci2016capacity,
     author = {Cayci, Semih and Eryilmaz, Atilla},
     title = {{On the Multi-Channel Capacity Gains of Millimeter-Wave Communication}},
     booktitle = {2016 IEEE Global Communications Conference (GLOBECOM)},
     pages = {1--6},
     year = {2016},
     publisher = {IEEE},
     doi = {10.1109/GLOCOM.2016.7842058},
     url = {https://doi.org/10.1109/GLOCOM.2016.7842058},
   }

2013

1. Lossless Polar Compression of q-ary Sources

   Semih Cayci and Orhan Arikan

   In 2013 IEEE International Symposium on Information Theory, 2013

   DOI Bib

   @inproceedings{cayci2013compression,
     author = {Cayci, Semih and Arikan, Orhan},
     title = {{Lossless Polar Compression of q-ary Sources}},
     booktitle = {2013 IEEE International Symposium on Information Theory},
     pages = {1132--1136},
     year = {2013},
     publisher = {IEEE},
     doi = {10.1109/ISIT.2013.6620403},
     url = {https://doi.org/10.1109/ISIT.2013.6620403},
   }

2012

1. Polar Code Construction for Non-Binary Source Alphabets

   Semih Cayci, Orhan Arikan, and Erdal Arikan

   In 2012 20th Signal Processing and Communications Applications Conference (SIU), 2012

   DOI Bib

   @inproceedings{cayci2012polar,
     author = {Cayci, Semih and Arikan, Orhan and Arikan, Erdal},
     title = {{Polar Code Construction for Non-Binary Source Alphabets}},
     booktitle = {2012 20th Signal Processing and Communications Applications Conference (SIU)},
     pages = {1--4},
     year = {2012},
     publisher = {IEEE},
     doi = {10.1109/SIU.2012.6204702},
     url = {https://doi.org/10.1109/SIU.2012.6204702},
   }

2026

1. A Comparative Analysis of Attention versus State-Space Models for In-Context Learning

   Enes Arda, Semih Cayci, and Atilla Eryilmaz

   2026

   Preprint; manuscript under review

   DOI HTML Bib

   @misc{arda2026attention,
     author = {Arda, Enes and Cayci, Semih and Eryilmaz, Atilla},
     title = {{A Comparative Analysis of Attention versus State-Space Models for In-Context Learning}},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {cs.LG},
     note = {Preprint; manuscript under review},
     doi = {10.48550/arXiv.2609.32341},
     url = {https://arxiv.org/abs/2609.32341},
   }

2. Preference-Based Reward Learning under Partial Observability with Inexact Dynamics

   Reza Zolnouri and Semih Cayci

   2026

   Preprint; manuscript under review

   DOI HTML Bib

   @misc{zolnouri2026preference,
     author = {Zolnouri, Reza and Cayci, Semih},
     title = {{Preference-Based Reward Learning under Partial Observability with Inexact Dynamics}},
     year = {2026},
     archiveprefix = {arXiv},
     primaryclass = {math.OC},
     note = {Preprint; manuscript under review},
     doi = {10.48550/arXiv.2606.30271},
     url = {https://arxiv.org/abs/2606.30271},
   }