Retrieve-then-Adapt: Retrieval-Augmented Test-Time Adaptation for Sequential Recommendation

2026-04-06

AI Summary

This work addresses the challenge that sequential recommendation models struggle to adapt to users’ evolving real-time preferences during inference due to distribution shifts and parameter constraints. To overcome this limitation, the authors propose ReAd, a test-time adaptation framework that requires no additional training. ReAd retrieves collaboratively similar items to construct enhanced embeddings and dynamically fuses them with predictions from a pretrained model through a lightweight adaptation module. By innovatively integrating collaborative memory retrieval with a low-overhead adaptation mechanism, ReAd achieves a superior balance between recommendation accuracy and adaptation efficiency. Extensive experiments on five benchmark datasets demonstrate that ReAd significantly outperforms existing methods, effectively breaking the trade-off between predictive performance and computational cost during inference.

Abstract

The sequential recommendation (SR) task aims to predict the next item based on users' historical interaction sequences. Typically trained on historical data, SR models often struggle to adapt to real-time preference shifts during inference due to challenges posed by distributional divergence and parameterized constraints. Existing approaches to address this issue include test-time training, test-time augmentation, and retrieval-augmented fine-tuning. However, these methods either introduce significant computational overhead, rely on random augmentation strategies, or require a carefully designed two-stage training paradigm. In this paper, we argue that the key to effective test-time adaptation lies in achieving both effective augmentation and efficient adaptation. To this end, we propose Retrieve-then-Adapt (ReAd), a novel framework that dynamically adapts a deployed SR model to the test distribution through retrieved user preference signals. Specifically, given a trained SR model, ReAd first retrieves collaboratively similar items for a test user from a constructed collaborative memory database. A lightweight retrieval learning module then integrates these items into an informative augmentation embedding that captures both collaborative signals and prediction-refinement cues. Finally, the initial SR prediction is refined via a fusion mechanism that incorporates this embedding. Extensive experiments across five benchmark datasets demonstrate that ReAd consistently outperforms existing SR methods.

Problem

• sequential recommendation
• test-time adaptation
• distributional divergence
• real-time preference shifts
• retrieval-augmented

Innovation

• Retrieve-then-Adapt
• retrieval-augmented adaptation
• test-time adaptation
• sequential recommendation
• collaborative memory