Privacy-preserving Model Upgrades With Bidirectional Compatible Training In Image Retrieval

Abstract

The task of privacy-preserving model upgrades in image retrieval desires to reap the benefits of rapidly evolving new models without accessing the raw gallery images. A pioneering work introduced backward-compatible training, where the new model can be directly deployed in a backfill-free manner, i.e., the new query can be directly compared to the old gallery features. Despite a possible solution, its improvement in sequential model upgrades is gradually limited by the fixed and under-quality old gallery embeddings. To this end, we propose a new model upgrade paradigm, termed Bidirectional Compatible Training (BiCT), which will upgrade the old gallery embeddings by forward-compatible training towards the embedding space of the backward-compatible new model. We conduct comprehensive experiments to verify the prominent improvement by BiCT and interestingly observe that the inconspicuous loss weight of backward compatibility actually plays an essential role for both backward and forward r

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keysu2022privacy

Related papers

• Boundary-aware Backward-compatible Representation Via Adversarial Learning In Image Retrieval (2023)9.21
• Metric Compatible Training For Online Backfilling In Large-scale Retrieval (2023)2.26
• Forward Compatible Training For Large-scale Embedding Retrieval Systems (2021)8.09
• Hot-refresh Model Upgrades With Regression-alleviating Compatible Training In Image Retrieval (2022)2.83
• Towards Backward-compatible Representation Learning (2020)12.93
• Darwinian Model Upgrades: Model Evolving With Selective Compatibility (2022)2.26
• Backward-compatible Aligned Representations Via An Orthogonal Transformation Layer (2024)0.00
• Learning Compatible Embeddings (2021)13.80