Melih Barsbey

5 papers · 2021–2025 · 5 venues · across top CS/AI venues

Authors are matched by name, so researchers who share a name ("Yang Liu", "Wei Wang") appear as a single profile and their paper counts, venues and achievements are combined. How identity is handled.

Achievements

Jump to papers ↓

🌍 Conference Polyglot · Silver 🌉 Interdisciplinary Bridge · Bronze 🗺️ Taxonomy Completionist (10)

Venues

ALT (1) ICCV (1) ICLR (1) ICML (1) NeurIPS (1)

Top co-authors

Umut Simsekli (2) Lucas Prieto (2) Tolga Birdal (2) Lingjiong Zhu (1) Stefanos Zafeiriou (1) Yijun Wan (1) Mert Gurbuzbalaban (1) Abdellatif Zaidi (1) Anant Raj (1) Pedro A. M. Mediano (1)

Research topics

Machine Learning (5) Optimization & Theory · Machine Learning (5) Deep Learning (2) Application Areas (2) Mathematics & Optimization (1) Architectures (1) Learning Types (1) Mathematics (1) Optimization & Theory · Deep Learning (1)

Keywords

stochastic gradient descent (2) heavy-tailed distribution (2) deep learning theory (1) neural network optimization (1) algorithmic stability (1) generalization error (1) learning rate (1) generalization bound (1) stochastic differential equation (1) spurious correlation (1) numerical stability (1) overparameterized network (1) least square (1) activation sparsity (1) euler discretization (1) heavy-tailed noise (1) large learning rate (1) invariant feature (1) network compression (1) heavy tail (1)

Papers

Grokking at the Edge of Numerical Stability ICLR 2025 Large Learning Rates Simultaneously Achieve Robustness to Spurious Correlations and Compressibility ICCV 2025 Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD ICML 2024 Algorithmic Stability of Heavy-Tailed Stochastic Gradient Descent on Least Squares ALT 2023 Heavy Tails in SGD and Compressibility of Overparametrized Neural Networks NeurIPS 2021