Leveraging Complementary Embeddings for Replay Selection in Continual Learning with Small Buffers
cs.LG
Submitted: 2026-04-09
Updated: 2026-09-16
License: http://creativecommons.org/licenses/by/4.0/
The gist: Catastrophic forgetting remains a key challenge in Continual Learning (CL).
Terminology
Abstract
Catastrophic forgetting remains a key challenge in Continual Learning (CL). In replay-based CL with severe memory constraints, performance critically depends on the sample selection strategy for the replay buffer. Most existing approaches construct memory buffers using embeddings learned under supervised objectives. However, class-agnostic, self-supervised representations often encode rich, class-relevant semantics that are overlooked. We propose a new method, Multiple Embedding Replay Selection, MERS, which replaces the buffer selection module with a graph-based approach that integrates both supervised and self-supervised embeddings. Empirical results show consistent improvements over SOTA selection strategies across a range of continual learning algorithms, with particularly strong gains in low-memory regimes. On CIFAR-100 and TinyImageNet, MERS outperforms single-embedding baselines without adding model parameters or increasing replay volume, making it a practical, drop-in enhancement for replay-based continual learning.
Sources
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
- Large sample analysis of the median heuristic
- Rainbow Memory: Continual Learning with a Memory of Diverse Samples
- VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning
- New Insights on Reducing Abrupt Representation Change in Online Continual Learning
- Do Pre-trained Models Benefit Equally in Continual Learning?
- Self-Supervised Class Incremental Learning
- Representation Learning with Contrastive Predictive Coding
- DINOv2: Learning Robust Visual Features without Supervision
- Overcoming catastrophic forgetting with hard attention to the task
- TEAL: New Selection Strategy for Small Buffers in Experience Replay Class Incremental Learning
- The information bottleneck method
- Lifelong Learning with Dynamically Expandable Networks
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks