Reproduce “Quantum Continual Learning Overcoming Catastrophic Forgetting” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Wenjie Jiang, Zhide Lu, Dong-Ling Deng
This paper investigates whether variational quantum classifiers suffer from catastrophic forgetting, confirming its presence across diverse tasks. It then adapts the classical Elastic Weight Consolidation (EWC) method to quantum continual learning, successfully preserving knowledge on prior tasks while learning new ones.
Key insights
- Catastrophic forgetting is not unique to classical neural networks; variational quantum classifiers also suffer from significant knowledge degradation when learning new tasks sequentially.
- Classical continual learning techniques, specifically EWC, can be successfully adapted to quantum machine learning to mitigate forgetting.
- The Fisher Information Matrix effectively quantifies the importance of variational parameters for a given task, providing a robust mechanism for regularization in quantum continual learning.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.