Batch Model Consolidation: A Multi-Task Model Consolidation Framework

In Continual Learning (CL), a model is required to learn a stream of tasks sequentially without significant performance degradation on previously learned tasks. Current approaches fail for a long sequence of tasks from diverse domains and difficulties. Many of the existing CL approaches are difficult to apply in practice due to excessive memory cost or training time, or are tightly coupled to a single device. With the intuition derived from the widely applied mini-batch training, we propose Batch Model Consolidation (BMC) to support more realistic CL under conditions where multiple agents are exposed to a range of tasks. During a textitregularization phase, BMC trains multiple textitexpert models in parallel on a set of disjoint tasks. Each expert maintains weight similarity to a textitbase model through a textitstability loss, and constructs a textitbuffer from a fraction of the task's data. During the textitconsolidation phase, we combine the learned knowledge on 'batches' of textitexpert models using a textitbatched consolidation loss in textitmemory data that aggregates all buffers. We thoroughly evaluate each component of our method in an ablation study and demonstrate the effectiveness on standardized benchmark datasets Split-CIFAR-100, Tiny-ImageNet, and the Stream dataset composed of 71 image classification tasks from diverse domains and difficulties. Our method outperforms the next best CL approach by 70% and is the only approach that can maintain performance at the end of 71 tasks; Our benchmark can be accessed at https://github.com/fostiropoulos/streambenchmark