Jul 2026
A Deterministic Binary Fingerprinting Framework with Zero-Trained Feature Extraction for Sparse Count Matrices
MMTB is introduced, a deterministic non-learned binary representation framework that requires no label supervision, model fitting, or gradient-based optimization and is suited for coarse-grained separation and resource-constrained deployments rather than fine-grained subtype discovery.
Lei Zhao, Fujin Huang, Ling Kang et al.
· arXiv.org · 0 citations