Variance & Greediness: A Comparative Study Of Metric-learning Losses

Abstract

Metric learning is central to retrieval, yet its effects on embedding geometry and optimization dynamics are not well understood. We introduce a diagnostic framework, VARIANCE (intra-/inter-class variance) and GREEDINESS (active ratio and gradient norms), to compare seven representative losses, i.e., Contrastive, Triplet, N-pair, InfoNCE, ArcFace, SCL, and CCL, across five image-retrieval datasets. Our analysis reveals that Triplet and SCL preserve higher within-class variance and clearer inter-class margins, leading to stronger top-1 retrieval in fine-grained settings. In contrast, Contrastive and InfoNCE compact embeddings are achieved quickly through many small updates, accelerating convergence but potentially oversimplifying class structures. N-pair achieves a large mean separation but with uneven spacing. These insights reveal a form of efficiency-granularity trade-off and provide practical guidance: prefer Triplet/SCL when diversity preservation and hard-sample discrimination are

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyzeng2026variance

Related papers

• Comparing Contrastive And Triplet Loss: Variance Analysis And Optimization Behavior (2025)0.00
• Rethinking Preventing Class-collapsing In Metric Learning With Margin-based Losses (2020)9.23
• Constellation Loss: Improving The Efficiency Of Deep Metric Learning Loss Functions For Optimal Embedding (2019)0.00
• Shadow Loss: Memory-linear Deep Metric Learning For Efficient Training (2023)0.00
• Deep Metric Learning Assisted By Intra-variance In A Semi-supervised View Of Learning (2023)5.24
• Do Lessons From Metric Learning Generalize To Image-caption Retrieval? (2022)5.24
• Metric Learning As A Service With Covariance Embedding (2022)3.58
• MIC: Mining Interclass Characteristics For Improved Metric Learning (2019)11.93