Reducing catastrophic forgetting with learning on synthetic data
This addresses a critical limitation for deep learning applications where data changes over time, though it is incremental as it builds on existing synthetic data approaches.
The paper tackled catastrophic forgetting in neural networks by generating synthetic data via a two-step optimization process with meta-gradients, resulting in no forgetting when trained sequentially on the Split-MNIST dataset.
Catastrophic forgetting is a problem caused by neural networks' inability to learn data in sequence. After learning two tasks in sequence, performance on the first one drops significantly. This is a serious disadvantage that prevents many deep learning applications to real-life problems where not all object classes are known beforehand; or change in data requires adjustments to the model. To reduce this problem we investigate the use of synthetic data, namely we answer a question: Is it possible to generate such data synthetically which learned in sequence does not result in catastrophic forgetting? We propose a method to generate such data in two-step optimisation process via meta-gradients. Our experimental results on Split-MNIST dataset show that training a model on such synthetic data in sequence does not result in catastrophic forgetting. We also show that our method of generating data is robust to different learning scenarios.