Randomised-pass replay improves continual learning accuracy without extra cost
Beyond Class Marginals: Bounding Rehearsal Gaps without Freezing Class Co-occurrence
Machine LearningArtificial Intelligence
Summary
In continual learning, how often a class is revisited during training affects how well an AI remembers it. The authors studied the timing between revisits, called the rehearsal gap, and introduced a method called randomised-pass replay (RPR) that controls this gap better without needing extra data or computation. They showed that RPR reduces bias in learning and improves accuracy on challenging tasks compared to standard balanced replay methods. These benefits depend on factors like the model used and how replay batches are sized.
What this means in practice
- •For machine learning engineers: Improve continual learning models by scheduling replay to limit maximum class absence intervals, boosting accuracy without added compute.
- •For computer vision developers: Enhance vision transformer-based models on sequential image tasks by using randomised-pass replay for small batch continual learning.
Authors
Congren Dai, Nat Roongjirarat, Fei Ye
Abstract
Class-balanced replay controls class frequency but does not determine the interval between successive replay appearances of a class. We study this interval, the rehearsal gap, separately from the class marginal and class co-occurrence, and introduce randomised-pass replay (RPR), which visits each resident class once per shuffled pass. For a fixed set of C resident classes and replay batch size b less than or equal to C, RPR preserves the balanced time-averaged class marginal and bounds every gap by 2*ceil(C/b)-1; a churn-conditional bound applies while the resident set changes. The scheduler uses no future class information and adds no replay examples or forward passes. In a linear-head ER-ACE diagnostic, joint absence from the incoming and replay batches produces a one-sided classifier-bias gradient. Longer absence episodes are associated with larger negative bias displacement, and removing the incoming-loss mask attenuates the scheduling effect. In the primary ER-ACE experiments, RPR improves final average accuracy by 0.72-1.67 percentage points relative to independent class-balanced retrieval under reservoir storage, with positive effects also observed under balanced storage. Pretrained ViTs show positive effects on the tested LT10 streams with small replay batches, while matched larger-batch controls show no material effect. Fixed-cycle and reused-pass controls change more than one temporal statistic, so the experiments do not isolate rehearsal-gap length from all other forms of temporal dependence. The accuracy effects depend on the learner and operating regime.