Tag: catastrophic forgetting

A new class-incremental learning framework called DARE-IDS uses diversity-aware replay exemplars to help intrusion detection systems learn new attack classes ...

Every time a traveler passes through an automated border control gate at a major international airport, a face verification system ...

A new hypothesis paper argues that the ordered stages of human neurodevelopment, from reflexes to physical grounding to social reasoning, ...

An international study of 30 neonatal intensive care units shows that continual learning can adapt adult-trained AI models to newborn ...

Researchers have unveiled a meta-learning model that recognizes ship types from underwater sound using only a few examples while retaining ...

Japanese researchers show that federated learning networks can progressively improve brain lesion detection software as hospitals join one at a ...

Researchers show that a randomly initialized, never-trained graph neural network paired with a lightweight streaming classifier outperforms state-of-the-art methods in ...

Researchers have developed a hypernetwork-based framework that lets continually learning AI systems erase specific tasks without access to the original ...

Penn State researchers have developed a continual learning method that adaptively adjusts LoRA ranks layer by layer, helping large language ...

Researchers in India have developed BiPC-IFS, a few-shot class-incremental learning framework that lets intrusion detection systems learn new cyberattacks from ...