Continual Learning with Fully Probabilistic Models
arXiv:2104.09240
Abstract
We present an approach for continual learning (CL) that is based on fully probabilistic (or generative) models of machine learning. In contrast to, e.g., GANs that are "generative" in the sense that they can generate samples, fully probabilistic models aim at modeling the data distribution directly. Consequently, they provide functionalities that are highly relevant for continual learning, such as density estimation (outlier detection) and sample generation. As a concrete realization of generative continual learning, we propose Gaussian Mixture Replay (GMR). GMR is a pseudo-rehearsal approach using a Gaussian Mixture Model (GMM) instance for both generator and classifier functionalities. Relying on the MNIST, FashionMNIST and Devanagari benchmarks, we first demonstrate unsupervised task boundary detection by GMM density estimation, which we also use to reject untypical generated samples. In addition, we show that GMR is capable of class-conditional sampling in the way of a cGAN. Lastly, we verify that GMR, despite its simple structure, achieves state-of-the-art performance on common class-incremental learning problems at very competitive time and memory complexity.
arXiv commentAccepted as Findings at the CLVISION2021 workshop, 11 pages, 6 figures
References in corpus (12)
- Overcoming catastrophic forgetting in neural networks
- Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
- A continual learning survey: Defying forgetting in classification tasks
- Continual Learning Through Synaptic Intelligence
- Efficient Lifelong Learning with A-GEM
- Expansions for nearly Gaussian distributions
- Overcoming Catastrophic Forgetting by Incremental Moment Matching
- Measuring Catastrophic Forgetting in Neural Networks
- Towards Robust Evaluations of Continual Learning
- Deep Generative Dual Memory Network for Continual Learning
- Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks
- A comprehensive, application-oriented study of catastrophic forgetting in DNNs