Class Incremental Learning for Algorithm Selection
This work addresses the incremental problem of enabling algorithm selection models to adapt to new data and solvers in streaming optimization, which is important for practitioners in optimization domains.
The paper tackles the problem of algorithm selection in streaming optimization scenarios where new instances and solvers arrive over time, requiring models to be updated without catastrophic forgetting. It benchmarks 8 continual learning methods on a bin-packing dataset, finding that rehearsal-based methods significantly outperform others with only about 7% forgetting loss.
Algorithm selection is commonly used to predict the best solver from a portfolio per per-instance. In many real scenarios, instances arrive in a stream: new instances become available over time, while the number of class labels can also grow as new data distributions arrive downstream. As a result, the classification model needs to be periodically updated to reflect additional solvers without catastrophic forgetting of past data. In machine-learning (ML), this is referred to as Class Incremental Learning (CIL). While commonly addressed in ML settings, its relevance to algorithm-selection in optimisation has not been previously studied. Using a bin-packing dataset, we benchmark 8 continual learning methods with respect to their ability to withstand catastrophic forgetting. We find that rehearsal-based methods significantly outperform other CIL methods. While there is evidence of forgetting, the loss is small at around 7%. Hence, these methods appear to be a viable approach to continual learning in streaming optimisation scenarios.