Mete Erdogan

I am a first-year PhD student in Electrical Engineering at Stanford University, currently rotating with Mert Pilanci and Chris Ré. My research focuses on the theoretical and algorithmic foundations of efficient, large-scale learning systems.

I am particularly interested in LLM compression, efficient sequence modeling, and alternative learning rules that reduce memory, compute, or optimization overhead. My work explores how structure in optimization, information theory, and signal processing can be leveraged to design scalable and adaptive learning mechanisms.

Before Stanford, I worked at EPFL with Volkan Cevher on language model pruning, compression, and spectral interpretability (e.g., PCA/CCA), and non-Euclidean optimization methods for deep learning. I completed dual B.S. degrees in Electrical and Electronics Engineering and Computer Science at Koç University, where I worked with Alper T. Erdogan on deep learning theory and biologically plausible learning rules alternative to backpropagation. I also collaborated with Deniz Yuret on developing Turkish language models, with Metin Sitti at the Max Planck Institute on learning-based modeling and control, and with Alper Demir on Kalman filtering for sensor tracking.

Email  /  GitHub  /  Google Scholar  /  LinkedIn

Research

I’m broadly interested in efficient learning and sequence modeling. Current themes include: (i) Efficient attention/memory mechanisms for long context, (ii) Model compression for LLMs and VLMs, and (iii) Convex/Non-Convex Optimization.

Publications

• Efficient Large Language Model Inference with Neural Block Linearization
  Mete Erdogan, Francesco Tonin, Volkan Cevher.
  NeurIPS 2025 (Main Track). arXiv
• Generalized Gradient Norm Clipping & Non-Euclidean (L0,L1)-Smoothness
  T. Pethick, W. Xie, Mete Erdogan, K. Antonakopoulos, T. Silveti-Falls, Volkan Cevher.
  NeurIPS 2025 (Oral). arXiv
• Error Broadcast and Decorrelation as a Potential Artificial and Natural Learning Mechanism
  Mete Erdogan, Cengiz Pehlevan, Alper T. Erdogan.
  NeurIPS 2025 (Spotlight). arXiv
• Bridging the Bosphorus: Leveraging Turkish Large Language Models through Strategies for Low-Resource Language Adaptation and Benchmarking
  E. Acikgoz, Mete Erdogan, Deniz Yuret.
  EMNLP 2024 MRL Workshop. arXiv
• Machine Learning and Kalman Filtering for Nanomechanical Mass Spectrometry
  Mete Erdogan, N. B. Baytekin, S. E. Coban, Alper Demir.
  IEEE Sensors Journal (2024). IEEE

Education