Darwinian Model Upgrades: Model Evolving with Selective Compatibility
This addresses the cost and efficiency issues in industrial-scale retrieval systems, offering an incremental improvement over existing backward-compatible methods.
The paper tackles the expensive backfilling problem in model upgrades for retrieval systems by proposing Darwinian Model Upgrades (DMU), which uses selective backward compatibility and forward adaptation to improve new-to-new performance and new-to-old compatibility, as demonstrated on large-scale benchmarks.
The traditional model upgrading paradigm for retrieval requires recomputing all gallery embeddings before deploying the new model (dubbed as "backfilling"), which is quite expensive and time-consuming considering billions of instances in industrial applications. BCT presents the first step towards backward-compatible model upgrades to get rid of backfilling. It is workable but leaves the new model in a dilemma between new feature discriminativeness and new-to-old compatibility due to the undifferentiated compatibility constraints. In this work, we propose Darwinian Model Upgrades (DMU), which disentangle the inheritance and variation in the model evolving with selective backward compatibility and forward adaptation, respectively. The old-to-new heritable knowledge is measured by old feature discriminativeness, and the gallery features, especially those of poor quality, are evolved in a lightweight manner to become more adaptive in the new latent space. We demonstrate the superiority of DMU through comprehensive experiments on large-scale landmark retrieval and face recognition benchmarks. DMU effectively alleviates the new-to-new degradation and improves new-to-old compatibility, rendering a more proper model upgrading paradigm in large-scale retrieval systems.