Catastrophic forgetting is the core problem of class incremental learning (CIL). Existing work
mainly adopts memory replay, knowledge distillation, and dynamic architecture to alleviate …

Class-incremental learning (CIL) aims to recognize new classes incrementally while
maintaining the discriminability of old classes. Most existing CIL methods are exemplar …

Existing research has shown that large language models (LLMs) exhibit remarkable
performance in language understanding and generation. However, when LLMs are …

Deep neural networks perform remarkably well in close-world scenarios. However, novel
classes emerged continually in real applications, making it necessary to learn incrementally …

Federated learning (FL) involves multiple heterogeneous clients collaboratively training a
global model via iterative local updates and model fusion. The generalization of FL's global …

Continual learning has recently become a primary paradigm for deep neural network
models in modern artificial intelligence services, where streaming data patterns frequently …

Entezari et al.(2022) conjectured that neural network solution sets reachable via stochastic
gradient descent (SGD) are convex, considering permutation invariances. This means that …

Few-shot class-incremental learning (FSCIL) aims at recognizing novel classes continually
with limited novel class samples. A mainstream baseline for FSCIL is first to train the whole …