Forward Compatible Training For Large-scale Embedding Retrieval Systems

Abstract

In visual retrieval systems, updating the embedding model requires recomputing features for every piece of data. This expensive process is referred to as backfilling. Recently, the idea of backward compatible training (BCT) was proposed. To avoid the cost of backfilling, BCT modifies training of the new model to make its representations compatible with those of the old model. However, BCT can significantly hinder the performance of the new model. In this work, we propose a new learning paradigm for representation learning: forward compatible training (FCT). In FCT, when the old model is trained, we also prepare for a future unknown version of the model. We propose learning side-information, an auxiliary feature for each sample which facilitates future updates of the model. To develop a powerful and flexible framework for model compatibility, we combine side-information with a forward transformation from old to new embeddings. Training of the new model is not modified, hence, its accura

Stats

• citations11
• S2 citations—
• github stars0
• HF likes0
• heat score8.09
• arxiv keyramanujan2021forward

Related papers

• Towards Backward-compatible Representation Learning (2020)12.93
• Privacy-preserving Model Upgrades With Bidirectional Compatible Training In Image Retrieval (2022)0.00
• Learning Compatible Embeddings (2021)13.80
• 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
• Fastfill: Efficient Compatible Model Update (2023)0.00
• Backward-compatible Aligned Representations Via An Orthogonal Transformation Layer (2024)0.00
• Towards Cross-modal Backward-compatible Representation Learning For Vision-language Models (2024)0.00