Continual learning algorithms strive to acquire new knowledge while preserving prior information. Often, these algorithms emphasise stability and restrict network updates upon learning new tasks. In many cases, such restrictions come at a cost to the model's plasticity, i.e. the model's ability to adapt to the requirements of a new task. But is all change detrimental? Here, we approach this question by proposing that activation spaces in neural networks can be decomposed into two subspaces: a readout range in which change affects prior tasks and a null space in which change does not alter prior performance. Based on experiments with this novel technique, we show that, indeed, not all activation change is associated with forgetting. Instead, only change in the subspace visible to the readout of a task can lead to decreased stability, while restricting change outside of this subspace is associated only with a loss of plasticity. Analysing various commonly used algorithms, we show that regularisation-based techniques do not fully disentangle the two spaces and, as a result, restrict plasticity more than need be. We expand our results by investigating a linear model in which we can manipulate learning in the two subspaces directly and thus causally link activation changes to stability and plasticity. For hierarchical, nonlinear cases, we present an approximation that enables us to estimate functionally relevant subspaces at every layer of a deep nonlinear network, corroborating our previous insights. Together, this work provides novel means to derive insights into the mechanisms behind stability and plasticity in continual learning and may serve as a diagnostic tool to guide developments of future continual learning algorithms that stabilise inference while allowing maximal space for learning.
Enhancing Plasticity for First Session Adaptation Continual Learning
LTCI, Télécom-Paris, Institut Polytechnique de Paris（Télécom-Paris 工程院）；University of Bergamo（贝加莫大学）；Inria at University Grenoble Alpes, LJK（格勒诺布尔阿尔卑斯大学）
KeepLoRA: Continual Learning with Residual Gradient Adaptation
School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院）；Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, China（教育部计算机网络与信息集成重点实验室）；School of Software Technology, Zhejiang University（浙江大学软件学院）
KeepLoRA++: Continual Learning with Layer-Scaled Residual Gradient Adaptation
School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院）；Key Laboratory of Computer Network and Information Integration, Southeast University, Ministry of Education（东南大学计算机网络和信息集成教育部重点实验室）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）
IF2Net: Innately Forgetting-Free Networks for Continual Learning
Huazhong University of Science and Technology（华中科技大学）；Wuhan University of Technology（武汉理工大学）；National University of Singapore（新加坡国立大学）；Nanyang Technological University（南洋理工大学）；Qatar University（卡塔尔大学）
Revisiting Weight Regularization for Low-Rank Continual Learning
State Key Laboratory of Human-Machine Hybrid Augmented Intelligence（人机混合增强智能国家重点实验室）；Institute of Artificial Intelligence and Robotics（人工智能与机器人研究院）；Computer Vision Center（计算机视觉中心）；Universitat Autònoma de Barcelona（巴塞罗那自治大学）；School of Instrument Science and Engineering（仪器科学与工程学院）；Bernoulli Institute（伯努利学院）；University of Groningen（格罗宁根大学）；School of Software Engineering（软件工程学院）
