Libin Zhu

According to our database, Libin Zhu authored at least 15 papers between 2020 and 2024.

Collaborative distances:

• Dijkstra number of five.
• Erdős number of four.

2024

Toward Understanding the Dynamics of Over-parameterized Neural Networks

Libin Zhu

PhD thesis, 2024

Emergence in non-neural models: grokking modular arithmetic via average gradient outer product.

Neil Mallinar, Daniel Beaglehole, Libin Zhu, Adityanarayanan Radhakrishnan, Parthe Pandit, Mikhail Belkin

CoRR, 2024

Catapults in SGD: spikes in the training loss and their impact on generalization through feature learning.

Libin Zhu, Chaoyue Liu, Adityanarayanan Radhakrishnan, Mikhail Belkin

Proceedings of the Forty-first International Conference on Machine Learning, 2024

Quadratic models for understanding catapult dynamics of neural networks.

Libin Zhu, Chaoyue Liu, Adityanarayanan Radhakrishnan, Mikhail Belkin

Proceedings of the Twelfth International Conference on Learning Representations, 2024

2023

Neural tangent kernel at initialization: linear width suffices.

Arindam Banerjee, Pedro Cisneros-Velarde, Libin Zhu, Mikhail Belkin

Proceedings of the Uncertainty in Artificial Intelligence, 2023

Restricted Strong Convexity of Deep Learning Models with Smooth Activations.

Arindam Banerjee, Pedro Cisneros-Velarde, Libin Zhu, Misha Belkin

Proceedings of the Eleventh International Conference on Learning Representations, 2023

2022

Restricted Strong Convexity of Deep Learning Models with Smooth Activations.

Arindam Banerjee, Pedro Cisneros-Velarde, Libin Zhu, Mikhail Belkin

CoRR, 2022

A note on Linear Bottleneck networks and their Transition to Multilinearity.

Libin Zhu, Parthe Pandit, Mikhail Belkin

CoRR, 2022

Quadratic models for understanding neural network dynamics.

Libin Zhu, Chaoyue Liu, Adityanarayanan Radhakrishnan, Mikhail Belkin

CoRR, 2022

Transition to Linearity of General Neural Networks with Directed Acyclic Graph Architecture.

Libin Zhu, Chaoyue Liu, Mikhail Belkin

CoRR, 2022

Transition to Linearity of Wide Neural Networks is an Emerging Property of Assembling Weak Models.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

CoRR, 2022

Transition to Linearity of General Neural Networks with Directed Acyclic Graph Architecture.

Libin Zhu, Chaoyue Liu, Misha Belkin

Proceedings of the Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processing Systems 2022, 2022

Transition to Linearity of Wide Neural Networks is an Emerging Property of Assembling Weak Models.

Chaoyue Liu, Libin Zhu, Misha Belkin

Proceedings of the Tenth International Conference on Learning Representations, 2022

2020

Toward a theory of optimization for over-parameterized systems of non-linear equations: the lessons of deep learning.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

CoRR, 2020

On the linearity of large non-linear models: when and why the tangent kernel is constant.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

Proceedings of the Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, 2020