#unsupervised-lifelong-learning

#unsupervised-lifelong-learning

One challenge of LifeHD is its constrained problem scale which limits effectiveness in unsupervised lifelong learning, suggesting leveraging pretrained models for more complex tasks.

Formulating the unsupervised lifelong learning problem driven by real-world IoT applications using streaming sources and class-incremental models.

Build a classification algorithm mapping X → Y through a learning protocol and evaluation with an iid dataset for periodic testing.