Continual Learning with Recursive Gradient Optimization

By Hao Liu, Huaping Liu. 2022

56 citations recorded by the source corpus.

Learning multiple tasks sequentially without forgetting previous knowledge, called Continual Learning(CL), remains a long-standing challenge for neural networks. Most existing methods rely on additional network capacity or data replay. In contrast, we introduce a novel approach which we refer to as Recursive Gradient Optimization(RGO). RGO is composed of an iteratively updated optimizer that modifies the gradient to minimize forgetting without data replay and a virtual Feature Encoding Layer(FEL) that represents different long-term structures with only task descriptors. Experiments demonstrate that RGO has significantly better performance on popular continual classification benchmarks when compared to the baselines and achieves new state-of-the-art performance on 20-split-CIFAR100(82.22%) and 20-split-miniImageNet(72.63%). With higher average accuracy than Single-Task Learning(STL), this method is flexible and reliable to provide continual learning capabilities for learning models that rely on gradient descent.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Optimization & Training, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Hang Su
2. Jun Zhu
3. Liyuan Wang
4. Xingxing Zhang
5. Junshan Zhang
6. René Vidal
7. Sen Lin
8. Li Yang

In Computer Vision and Pattern Recognition

1. Hang Su
2. Jun Zhu
3. Liyuan Wang
4. Xingxing Zhang
5. René Vidal
6. Sen Lin
7. Li Yang
8. Paris V. Giampouras

In Optimization & Training

1. Liyuan Wang
2. René Vidal
3. Jun Zhu
4. Sen Lin
5. Hang Su
6. Li Yang
7. Paris V. Giampouras
8. Xingxing Zhang

In Meta & Continual Learning

1. Xingxing Zhang
2. Liyuan Wang
3. Hang Su
4. Jun Zhu
5. René Vidal
6. Sen Lin
7. Li Yang
8. Paris V. Giampouras

Citations in the corpus

• A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, methodology): “Under the upper bound of the quadratic penalty in Bayesian weight regularization, RGO [180] modifies gradient directions with a recursive optimization procedure to obtain the optimal solution.”
• The Ideal Continual Learner: An Agent That Never Forgets (2023, background): “…embody either of these three ideas, or combinations thereof, have greatly improved the empirical performance in resisting catastrophic forgetting (Liu, 2022).1 While these methods have evolved so rapidly, at least two questions remain under-explored: (Q1) Is there any mathematical connection…”
• Beyond Not-Forgetting: Continual Learning with Backward Knowledge Transfer (2022, methodology): “, [12, 1, 18]) penalize the modification of important weights of old tasks; parameter-isolation based methods (e.”