Abstract:Deep learning models suffer from catastrophic forgetting when learning new tasks incrementally. Incremental learning has been proposed to retain the knowledge of old classes while learning to identify new classes. A typical approach is to use a few exemplars to avoid forgetting old knowledge. In such a scenario, data imbalance between old and new classes is a key issue that leads to performance degradation of the model. Several strategies have been designed to rectify the bias towards the new classes due to data imbalance. However, they heavily rely on the assumptions of the bias relation between old and new classes. Therefore, they are not suitable for complex real-world applications. In this study, we propose an assumption-agnostic method, Multi-Granularity Regularized re-Balancing (MGRB), to address this problem. Re-balancing methods are used to alleviate the influence of data imbalance; however, we empirically discover that they would under-fit new classes. To this end, we further design a novel multi-granularity regularization term that enables the model to consider the correlations of classes in addition to re-balancing the data. A class hierarchy is first constructed by ontology or grouping semantically or visually similar classes. The multi-granularity regularization then transforms the one-hot label vector into a continuous label distribution, which reflects the relations between the target class and other classes based on the constructed class hierarchy. Thus, the model can learn the inter-class relational information, which helps enhance the learning of both old and new classes. Experimental results on both public datasets and a real-world fault diagnosis dataset verify the effectiveness of the proposed method. Code is available at https://github.com/lilyht/CIL-MGRB.

Mutually Promoted Hierarchical Learning for Incremental Implicitly-Refined Classification

Uncertainty-Aware Hierarchical Refinement for Incremental Implicitly-Refined Classification

IIRC: Incremental Implicitly-Refined Classification

Multi-Teacher Knowledge Distillation for Incremental Implicitly-Refined Classification

HCV: Hierarchy-Consistency Verification for Incremental Implicitly-Refined Classification

Incremental Classification of Remote Sensing Images with Feature Pyramid and Class Hierarchy

Multi-Granularity Regularized Re-Balancing for Class Incremental Learning

Incremental Learning of Remote Sensing Target Classification with Class Hierarchy

Memory Efficient Class-Incremental Learning for Image Classification

A Class Incremental Learning Method with Weight Re-Initialization and Multiple Distillation for Image Classification

Rebalancing network with knowledge stability for class incremental learning

Effective Decision Boundary Learning for Class Incremental Learning

Dynamic Residual Classifier for Class Incremental Learning

Class Incremental Learning with Deep Contrastive Learning and Attention Distillation

An integrated classification model for incremental learning

Distillation-Constrained Prototype Representation Network for Hyperspectral Image Incremental Classification

Learning to Classify With Incremental New Class

Self-Paced Imbalance Rectification for Class Incremental Learning

General Incremental Learning with Domain-aware Categorical Representations

Few-Shot Class-Incremental Learning Via Class-Aware Bilateral Distillation

Rethinking Class-Incremental Learning from a Dynamic Imbalanced Learning Perspective