DeepAI AI Chat
Log In Sign Up

Gradient Starvation: A Learning Proclivity in Neural Networks

by   Mohammad Pezeshki, et al.

We identify and formalize a fundamental gradient descent phenomenon resulting in a learning proclivity in over-parameterized neural networks. Gradient Starvation arises when cross-entropy loss is minimized by capturing only a subset of features relevant for the task, despite the presence of other predictive features that fail to be discovered. This work provides a theoretical explanation for the emergence of such feature imbalance in neural networks. Using tools from Dynamical Systems theory, we identify simple properties of learning dynamics during gradient descent that lead to this imbalance, and prove that such a situation can be expected given certain statistical structure in training data. Based on our proposed formalism, we develop guarantees for a novel regularization method aimed at decoupling feature learning dynamics, improving accuracy and robustness in cases hindered by gradient starvation. We illustrate our findings with simple and real-world out-of-distribution (OOD) generalization experiments.


Learning Unstable Dynamical Systems with Time-Weighted Logarithmic Loss

When training the parameters of a linear dynamical model, the gradient d...

On the Learning Dynamics of Deep Neural Networks

While a lot of progress has been made in recent years, the dynamics of l...

Persistency of Excitation for Robustness of Neural Networks

When an online learning algorithm is used to estimate the unknown parame...

Rapid Feature Evolution Accelerates Learning in Neural Networks

Neural network (NN) training and generalization in the infinite-width li...

Imitating Deep Learning Dynamics via Locally Elastic Stochastic Differential Equations

Understanding the training dynamics of deep learning models is perhaps a...

On how to avoid exacerbating spurious correlations when models are overparameterized

Overparameterized models fail to generalize well in the presence of data...