LoomSC: Scalable Deep Subspace Clustering with Projector Factorization and Exact Spectral Reduction

Key points

• Dense self-expression matrices and full-affinity spectral clustering limit the scalability of subspace clustering.1
• Motivated by the spectral structure of least-squares regression, LoomSC jointly learns latent features and a projector self-representation through two thin factors.1
• An exact feature map then reduces its normalized spectral problem to an eigenproblem whose dimension depends only on the factor width.1
• Our analysis quantifies the projector approximation and identifies conditions for subspace preservation and within-subspace connectivity.1

Sources (1)

• [1]LoomSC: Scalable Deep Subspace Clustering with Projector Factorization and Exact Spectral Reduction
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 03:37 PM

  We introduce the Latent Orthogonal Optimization Model for Subspace Clustering (LoomSC), a framework that addresses both bottlenecks through projector factorization and exact spectral reduction.

  Dense self-expression matrices and full-affinity spectral clustering limit the scalability of subspace clustering.

Extractive summary: sentences quoted from the sources.