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.

Large Scale Incremental Learning

Class-Incremental Learning: A Survey

Class-Incremental Learning: Survey and Performance Evaluation on Image Classification

Maintaining Discrimination and Fairness in Class Incremental Learning

Learning a Unified Classifier Incrementally via Rebalancing

Class Incremental Learning Based on Local Structure Constraints in Feature Space

Rethinking Class-Incremental Learning from a Dynamic Imbalanced Learning Perspective

Incremental Learning In Online Scenario

Class-Incremental Learning Based on Big Dataset Pre-Trained Models

Discrimination Correction and Balance for Class-Incremental Learning

Multi-Granularity Regularized Re-Balancing for Class Incremental Learning

Class Incremental Learning with Deep Contrastive Learning and Attention Distillation

Error-Driven Incremental Learning in Deep Convolutional Neural Network for Large-Scale Image Classification

Incremental Learning Using Conditional Adversarial Networks

An Incremental One Class Learning Framework for Large Scale Data.

Learning a Dual-Branch Classifier for Class Incremental Learning

A Strategy for an Uncompromising Incremental Learner

Incremental Learning Via Rate Reduction

A Broad Neural Network Structure for Class Incremental Learning.

Class Incremental Learning with Important and Diverse Memory.

Class-Incremental Learning for Semantic Segmentation in Aerial Imagery Via Distillation in All Aspects