A comprehensive, application-oriented study of catastrophic forgetting in DNNs
This work addresses the problem of catastrophic forgetting for researchers and practitioners in machine learning, but it is incremental as it builds on existing protocols and models.
The study tackled catastrophic forgetting in deep neural networks by conducting a large-scale empirical analysis under application-oriented conditions, finding that no model could avoid forgetting across all datasets and tasks.
We present a large-scale empirical study of catastrophic forgetting (CF) in modern Deep Neural Network (DNN) models that perform sequential (or: incremental) learning. A new experimental protocol is proposed that enforces typical constraints encountered in application scenarios. As the investigation is empirical, we evaluate CF behavior on the hitherto largest number of visual classification datasets, from each of which we construct a representative number of Sequential Learning Tasks (SLTs) in close alignment to previous works on CF. Our results clearly indicate that there is no model that avoids CF for all investigated datasets and SLTs under application conditions. We conclude with a discussion of potential solutions and workarounds to CF, notably for the EWC and IMM models.