What and How: Generalized Lifelong Spectral Clustering via Dual Memory
Gan Sun, Yang Cong, Jiahua Dong +3 more
Spectral clustering (SC) has become one of the most widely-adopted clustering algorithms, and been successfully applied into various applications. We in this work explore the problem of spectral clustering in a lifelong learning framework termed as <u>G</u> eneralized <u>L</u> ife <u>l</u> ong <u>S</u> pectral <u>C</u> lustering (GL <inline-formula><tex-math notation="LaTeX">$^2$</tex-math></inline-formula> SC). Different from most current studies, which concentrate on a fixed spectral clustering task set and cannot efficiently incorporate a new clustering task, the goal of our work is to establish a generalized model for new spectral clustering tasks by “What” and “How” to lifelong learn from past tasks. In respect of “what to lifelong learn”, our GL <inline-formula><tex-math notation="LaTeX">$^2$</tex-math></inline-formula> SC framework contains a dual memory mechanism with a deep…