Coordinate heterogeneity governs binary quantization performance via closed-form ranking fidelity expressions and a two-parameter scaling law, validated on 13 datasets across 6 embedding families.
International Conference on Machine Learning , pages=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Coordinate Heterogeneity Governs Binary Quantization: From InfoNCE to Recall
Coordinate heterogeneity governs binary quantization performance via closed-form ranking fidelity expressions and a two-parameter scaling law, validated on 13 datasets across 6 embedding families.