Search papers, labs, and topics across Lattice.
This paper introduces GRACE, a dynamic scaling framework for Class Incremental Learning (CIL) that adaptively manages model capacity using a cyclic "GRow, Assess, ComprEss" strategy. GRACE incorporates a saturation assessment phase to evaluate model capacity utilization, guiding decisions to either expand the architecture or compress backbones, thus preventing uncontrolled parameter growth. Experiments on CIL benchmarks show GRACE achieves state-of-the-art performance while reducing memory footprint by up to 73% compared to purely expansionist models.
Forget uncontrolled parameter growth in class incremental learning: GRACE adaptively scales model capacity, achieving state-of-the-art performance with a 73% memory reduction.
Class Incremental Learning (CIL) poses a fundamental challenge: maintaining a balance between the plasticity required to learn new tasks and the stability needed to prevent catastrophic forgetting. While expansion-based methods effectively mitigate forgetting by adding task-specific parameters, they suffer from uncontrolled architectural growth and memory overhead. In this paper, we propose a novel dynamic scaling framework that adaptively manages model capacity through a cyclic "GRow, Assess, ComprEss" (GRACE) strategy. Crucially, we supplement backbone expansion with a novel saturation assessment phase that evaluates the utilization of the model's capacity. This assessment allows the framework to make informed decisions to either expand the architecture or compress the backbones into a streamlined representation, preventing parameter explosion. Experimental results demonstrate that our approach achieves state-of-the-art performance across multiple CIL benchmarks, while reducing memory footprint by up to a 73% compared to purely expansionist models.
