Critical Points of Neural Networks: Analytical Forms and Landscape Properties
This work addresses the theoretical understanding of loss landscapes for neural networks, which is incremental but important for optimization convergence in machine learning.
The paper provides a full analytical characterization of critical points and global minimizers for square loss functions in various neural networks, showing that linear networks have no spurious local minima, while one-hidden-layer ReLU networks do.
Due to the success of deep learning to solving a variety of challenging machine learning tasks, there is a rising interest in understanding loss functions for training neural networks from a theoretical aspect. Particularly, the properties of critical points and the landscape around them are of importance to determine the convergence performance of optimization algorithms. In this paper, we provide full (necessary and sufficient) characterization of the analytical forms for the critical points (as well as global minimizers) of the square loss functions for various neural networks. We show that the analytical forms of the critical points characterize the values of the corresponding loss functions as well as the necessary and sufficient conditions to achieve global minimum. Furthermore, we exploit the analytical forms of the critical points to characterize the landscape properties for the loss functions of these neural networks. One particular conclusion is that: The loss function of linear networks has no spurious local minimum, while the loss function of one-hidden-layer nonlinear networks with ReLU activation function does have local minimum that is not global minimum.