Learning Compatible Embeddings

Abstract

Achieving backward compatibility when rolling out new models can highly reduce costs or even bypass feature re-encoding of existing gallery images for in-production visual retrieval systems. Previous related works usually leverage losses used in knowledge distillation which can cause performance degradations or not guarantee compatibility. To address these issues, we propose a general framework called Learning Compatible Embeddings (LCE) which is applicable for both cross model compatibility and compatible training in direct/forward/backward manners. Our compatibility is achieved by aligning class centers between models directly or via a transformation, and restricting more compact intra-class distributions for the new model. Experiments are conducted in extensive scenarios such as changes of training dataset, loss functions, network architectures as well as feature dimensions, and demonstrate that LCE efficiently enables model compatibility with marginal sacrifices of accuracies. The

Stats

• citations26
• S2 citations—
• github stars33
• HF likes0
• heat score13.80
• arxiv keymeng2021learning

Related papers

• Towards Backward-compatible Representation Learning (2020)12.93
• Forward Compatible Training For Large-scale Embedding Retrieval Systems (2021)8.09
• Unified Representation Learning For Cross Model Compatibility (2020)5.24
• Privacy-preserving Model Upgrades With Bidirectional Compatible Training In Image Retrieval (2022)0.00
• Stationary Representations: Optimally Approximating Compatibility And Implications For Improved Model Replacements (2024)5.87
• Metric Compatible Training For Online Backfilling In Large-scale Retrieval (2023)2.26
• Boundary-aware Backward-compatible Representation Via Adversarial Learning In Image Retrieval (2023)9.21
• Backward-compatible Aligned Representations Via An Orthogonal Transformation Layer (2024)0.00