Get started
Keyword

flat minima

Description: Broad, relatively level regions of an objective function's landscape, studied in machine learning for their relation to model generalization.
OpenAlex ID: keywords/flat-minima
GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
2017 · Heusel, Martin, Ramsauer, Hubert, et al. · arXiv (Cornell University) · 8,365
PDF
On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima
2016 · Keskar, Nitish Shirish, Mudigere, Dheevatsa, et al. · arXiv (Cornell University) · 1,128
PDF
Flat Minima
1997 · Sepp Hochreiter, Jürgen Schmidhuber · Neural Computation · 692
Averaging Weights Leads to Wider Optima and Better Generalization
2018 · Izmailov, Pavel, Podoprikhin, Dmitrii, et al. · arXiv (Cornell University) · 568
PDF
GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
2019 · Naresh Babu Bynagari · Asian Journal of Applied Science and Engineering · 513
PDF
Sharp Minima Can Generalize For Deep Nets
2017 · Dinh, Laurent, Pascanu, Razvan, et al. · arXiv (Cornell University) · 369
PDF
GANs Trained by a Two Time-Scale Update Rule Converge to a Nash Equilibrium
2017 · Martin Heusel, Hubert Ramsauer, et al. · arXiv (Cornell University) · 364
Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Data
2017 · Dziugaite, Gintare Karolina, Roy, Daniel M. · arXiv (Cornell University) · 320
PDF
Entropy-SGD: biasing gradient descent into wide valleys*
2019 · Pratik Chaudhari, Anna Choromanska, et al. · Journal of Statistical Mechanics Theory and Experiment · 293
Emergence of Invariance and Disentanglement in Deep Representations
2018 · Alessandro Achille, Stefano Soatto · Information Theory and Applications Workshop (ITA) · 293
PDF
SWAD: Domain Generalization by Seeking Flat Minima
2021 · Cha, Junbum, Chun, Sanghyuk, et al. · arXiv (Cornell University) · 209
PDF
Essentially No Barriers in Neural Network Energy Landscape
2018 · Draxler, Felix, Veschgini, Kambis, et al. · arXiv (Cornell University) · 203
PDF
Empirical Analysis of the Hessian of Over-Parametrized Neural Networks
2017 · Sagun, Levent, Evci, Utku, et al. · arXiv (Cornell University) · 153
PDF
Towards Theoretically Understanding Why SGD Generalizes Better Than ADAM in Deep Learning
2020 · Zhou, Pan, Feng, Jiashi, et al. · arXiv (Cornell University) · 146
PDF
Entropy-SGD: Biasing Gradient Descent Into Wide Valleys
2016 · Chaudhari, Pratik, Choromanska, Anna, et al. · arXiv (Cornell University) · 140
PDF
Sharpness-Aware Gradient Matching for Domain Generalization
2023 · Pengfei Wang, Zhaoxiang Zhang, et al. · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · 125
PDF
Towards Stable Test-Time Adaptation in Dynamic Wild World
2023 · Niu, Shuaicheng, Wu, Jiaxiang, et al. · arXiv (Cornell University) · 113
PDF
The Anisotropic Noise in Stochastic Gradient Descent: Its Behavior of Escaping from Sharp Minima and Regularization Effects
2018 · Zhu, Zhanxing, Wu, Jingfeng, et al. · ePrints Soton (University of Southampton) · 110
PDF
Biologically inspired protection of deep networks from adversarial attacks
2017 · Nayebi, Aran, Ganguli, Surya · arXiv (Cornell University) · 96
PDF
Hessian-based Analysis of Large Batch Training and Robustness to Adversaries
2018 · Yao, Zhewei, Gholami, Amir, et al. · arXiv (Cornell University) · 95
PDF
Improving Generalization in Federated Learning by Seeking Flat Minima
2022 · Debora Caldarola, Barbara Caputo, et al. · Lecture notes in computer science · 92
PDF
Overcoming Catastrophic Forgetting in Incremental Few-Shot Learning by Finding Flat Minima
2021 · Shi, Guangyuan, Chen, Jiaxin, et al. · arXiv (Cornell University) · 92
PDF
Simplifying Neural Nets by Discovering Flat Minima
1994 · Sepp Hochreiter, Jürgen Schmidhuber · 87
Do We Need Zero Training Loss After Achieving Zero Training Error?
2020 · Ishida, Takashi, Yamane, Ikko, et al. · arXiv (Cornell University) · 79
PDF
ps-CALR: Periodic-Shift Cosine Annealing Learning Rate for Deep Neural Networks
2023 · Olanrewaju Victor Johnson, Chew Xinying, et al. · IEEE Access · 70
PDF
Works count: 567
Citation count: 17,340
Year
Topic
Type