🤖 AI Summary
Deep neural networks often suffer from loss of plasticity in continual learning due to neuron saturation and unbounded growth of parameter norms, hindering effective acquisition of new tasks. This work proposes the Gradient versus Reference State Discrepancy (GXD) method, which formulates adaptive resetting as an intervention cost estimation problem for the first time. By leveraging reference-based gradient attribution and a first-order Taylor expansion, GXD precisely quantifies the functional cost of resetting individual neurons, enabling identification of inefficient units and guiding adaptive reinitialization. Experimental results demonstrate that GXD significantly outperforms existing activation- or gradient-based proxy methods across diverse continual learning scenarios, effectively restoring model plasticity and learning capacity.
📝 Abstract
Continual learning research attempts to conserve two fundamental capabilities: new knowledge acquisition and the preservation of previously acquired knowledge. While knowledge in this case can be measured through performance over an implicit or explicit task space, model plasticity generally concerns adaptability as data distributions evolve. Though much of the literature has focused on catastrophic forgetting, deep networks can also suffer from loss of plasticity, becoming progressively harder to update under continued training. Recent research has identified multiple mechanisms underlying this phenomenon, including neuron saturation, parameter norm growth, and loss of useful curvature directions. Adaptive reset-based interventions, which selectively reinitialize low-utility network parameters, have emerged as practical solutions to restore trainability. Existing utility measures used to guide resets, such as activation magnitude, contribution utility, or gradient-based activity, rely on proxy signals that can become misaligned with the intervention they are meant to guide. In this paper, we introduce gradient times difference from reference (GXD), a theoretically motivated utility measure based on reference-based gradient attribution that estimates the first-order functional cost of replacing a unit. Our results show that utility measures aligned with the functional cost of the reset can make interventions more reliable in settings where existing reset criteria degrade. GXD reframes adaptive resetting as an intervention cost estimation problem, providing a practical path toward more robust continual learning systems.