Deep Metric Learning with Chance Constraints
Not on the map. Why?
Abstract (arXiv)
Deep metric learning (DML) aims to minimize empirical expected loss of the pairwise intra-/inter- class proximity violations in the embedding space. We relate DML to feasibility problem of finite chance constraints. We show that minimizer of proxy-based DML satisfies certain chance constraints, and that the worst case generalization performance of the proxy-based methods can be characterized by the radius of the smallest ball around a class proxy to cover the entire domain of the corresponding class samples, suggesting multiple proxies per class helps performance. To provide a scalable algorithm as well as exploiting more proxies, we consider the chance constraints implied by the minimizers of proxy-based DML instances and reformulate DML as finding a feasible point in intersection of such constraints, resulting in a problem to be approximately solved by iterative projections. Simply put, we repeatedly train a regularized proxy-based loss and re-initialize the proxies with the embeddings of the deliberately selected new samples. We applied our method with 4 well-accepted DML losses and show the effectiveness with extensive evaluations on 4 popular DML benchmarks. Code is available at: https://github.com/yetigurbuz/ccp-dml
Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 49 of its 89 references that are in the map. First posted on arXiv 19 Sep 2022.
Builds on
49 of 89 references.
References · 49
- 2014Adam: A Method for Stochastic Optimizationcited by 36,960
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
- 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
- 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,833
- 2015Learning Deep Features for Discriminative Localizationcited by 2,430
- 2016Identity Mappings in Deep Residual Networkscited by 1,960
- 2015Deep Metric Learning via Lifted Structured Feature Embeddingcited by 449
- 2019Multi-Similarity Loss with General Pair Weighting for Deep Metric Learningcited by 368
- 2020A Metric Learning Reality Checkcited by 317
- 2017Sampling Matters in Deep Embedding Learningcited by 285
- 2017No Fuss Distance Metric Learning using Proxiescited by 225
- 2020Proxy Anchor Loss for Deep Metric Learningcited by 195
37 more
- 2019SoftTriple Loss: Deep Metric Learning Without Triplet Samplingcited by 163
- 2019Cross-Batch Memory for Embedding Learningcited by 130
- 2018Deep Metric Learning with Hierarchical Triplet Losscited by 112
- 2020ProxyNCA++: Revisiting and Revitalizing Proxy Neighborhood Component Analysiscited by 107
- 2020Revisiting Training Strategies and Generalization Performance in Deep Metric Learningcited by 99
- 2017Smart Mining for Deep Metric Learningcited by 96
- 2019Ranked List Loss for Deep Metric Learningcited by 77
- 2018Attention-based Ensemble for Deep Metric Learningcited by 56
- 2019Hardness-Aware Deep Metric Learningcited by 55
- 2021Recall@k Surrogate Loss with Large Batches and Similarity Mixupcited by 49
- 2016Hard-Aware Deeply Cascaded Embeddingcited by 44
- 2019MIC: Mining Interclass Characteristics for Improved Metric Learningcited by 41
- 2015Metric Learning with Adaptive Density Discriminationcited by 40
- 2020DiVA: Diverse Visual Feature Aggregation for Deep Metric Learningcited by 40
- 2019Divide and Conquer the Embedding Space for Metric Learningcited by 39
- 2020Fewer is More: A Deep Graph Metric Learning Perspective Using Fewer Proxiescited by 36
- 2021Learning Intra-Batch Connections for Deep Metric Learningcited by 34
- 2022Non-isotropy Regularization for Proxy-based Deep Metric Learningcited by 33
- 2021Towards Interpretable Deep Metric Learning with Structural Matchingcited by 30
- 2019Metric Learning With HORDE: High-Order Regularizer for Deep Embeddingscited by 29
- 2020Embedding Expansion: Augmentation in Embedding Space for Deep Metric Learningcited by 27
- 2019A Theoretically Sound Upper Bound on the Triplet Loss for Improving the Efficiency of Deep Distance Metric Learningcited by 25
- 2020Deep Metric Learning with Spherical Embeddingcited by 25
- 2021Noise-resistant Deep Metric Learning with Ranking-based Instance Selectioncited by 23
- 2021It Takes Two to Tango: Mixup for Deep Metric Learningcited by 23
- 2021Deep Relational Metric Learningcited by 22
- 2021Learning with Memory-based Virtual Classes for Deep Metric Learningcited by 21
- 2021Hierarchical Proxy-based Loss for Deep Metric Learningcited by 18
- 2018Deep Randomized Ensembles for Metric Learningcited by 17
- 2021Proxy Synthesis: Learning with Synthetic Classes for Deep Metric Learningcited by 17
- 2020S2SD: Simultaneous Similarity-based Self-Distillation for Deep Metric Learningcited by 14
- 2018Virtual Class Enhanced Discriminative Embedding Learningcited by 13
- 2019Unbiased Evaluation of Deep Metric Learning Algorithmscited by 10
- 2021Multi-level Distance Regularization for Deep Metric Learningcited by 6
- 2021Multi-Head Deep Metric Learning Using Global and Local Representationscited by 4
- 2019Deep Metric Learning with Alternating Projections onto Feasible Setscited by 3
- 2015Nonconvex set intersection problems: From projection methods to the Newton method for super-regular setscited by 1
Built on by
Citing papers per year: 2023: 2; 2024: 0; 2025: 0; 2026: 1.
3 papers.