Distribution and expression aware retrospective learning for single-cell long-tailed class-incremental annotation

doi:10.1016/j.asoc.2026.114745

Distribution and expression aware retrospective learning for single-cell long-tailed class-incremental annotation

2026-02-05

0

PRE

AI

张永清 (Yongqing Zhang) *

DOI：10.1016/j.asoc.2026.114745

Original

Original request for help

Save

Abstract

En 中文

• Proposed scLTCIA, a replay-based framework for long-tailed incremental annotation. • Introduced distribution-aware diffusion to accurately recover sparse scRNA-seq data. • Designed expression-aware distillation to align views and reduce gene forgetting. • Employed fuzzy-guided constraints to reduce spurious recall on novel cell types.

Keywords:

scLTCIA

long-tailed incremental annotation

distribution-aware diffusion

expression-aware distillation

fuzzy-guided constraints

Distribution and expression aware retrospective learning for single-cell long-tailed class-incremental annotation

Abstract

Journal

Organization