An Empirical Study of Example Forgetting during Deep Neural Network Learning
This addresses the problem of understanding learning dynamics and data efficiency for researchers in machine learning, though it is incremental as it builds on prior work on forgetting.
The study investigated whether neural networks forget specific training examples during learning on single tasks, finding that some examples are frequently forgotten while others are not, and that omitting a significant fraction of examples based on forgetting dynamics can maintain state-of-the-art generalization performance.
Inspired by the phenomenon of catastrophic forgetting, we investigate the learning dynamics of neural networks as they train on single classification tasks. Our goal is to understand whether a related phenomenon occurs when data does not undergo a clear distributional shift. We define a `forgetting event' to have occurred when an individual training example transitions from being classified correctly to incorrectly over the course of learning. Across several benchmark data sets, we find that: (i) certain examples are forgotten with high frequency, and some not at all; (ii) a data set's (un)forgettable examples generalize across neural architectures; and (iii) based on forgetting dynamics, a significant fraction of examples can be omitted from the training data set while still maintaining state-of-the-art generalization performance.