| Loss landscapes and optimization in over-parameterized non-linear systems and neural networks C Liu, L Zhu, M Belkin Applied and Computational Harmonic Analysis 59, 85-116, 2022 | 259* | 2022 |
| On the linearity of large non-linear models: when and why the tangent kernel is constant C Liu, L Zhu, M Belkin Advances in Neural Information Processing Systems 33, 15954-15964, 2020 | 140 | 2020 |
| Quadratic models for understanding neural network dynamics L Zhu, C Liu, A Radhakrishnan, M Belkin arXiv preprint arXiv:2205.11787, 2022 | 13 | 2022 |
| Catapults in sgd: spikes in the training loss and their impact on generalization through feature learning L Zhu, C Liu, A Radhakrishnan, M Belkin arXiv preprint arXiv:2306.04815, 2023 | 6 | 2023 |
| Transition to linearity of general neural networks with directed acyclic graph architecture L Zhu, C Liu, M Belkin Advances in neural information processing systems 35, 5363-5375, 2022 | 5 | 2022 |
| Restricted strong convexity of deep learning models with smooth activations A Banerjee, P Cisneros-Velarde, L Zhu, M Belkin arXiv preprint arXiv:2209.15106, 2022 | 5 | 2022 |
| Transition to linearity of wide neural networks is an emerging property of assembling weak models C Liu, L Zhu, M Belkin arXiv preprint arXiv:2203.05104, 2022 | 3 | 2022 |
| Neural tangent kernel at initialization: linear width suffices A Banerjee, P Cisneros-Velarde, L Zhu, M Belkin Uncertainty in Artificial Intelligence, 110-118, 2023 | 2 | 2023 |
| A note on Linear Bottleneck networks and their Transition to Multilinearity L Zhu, P Pandit, M Belkin arXiv preprint arXiv:2206.15058, 2022 | | 2022 |
Mikhail (Misha) BelkinProfessor of Data Science, Halıcıoğlu Data Science Institute, CSE, UCSD, Amazon ScholarVerified email at ucsd.edu
