Quantum Incremental Learning with Mixed State Prototypes

A new quantum incremental learning framework addresses circuit width limitations in NISQ devices by using trainable mixed-state prototypes instead of increasing backbone parameters. This method enables sequential class addition by expanding the prototype set, offering higher representation capacity than single pure-state prototypes for mitigating catastrophic forgetting.