Dual Test-time Training for Out-of-distribution Recommender System

TL;DR

This paper proposes a novel Dual Test-Time-Training framework for OOD Recommendation, termed DT3OR, which is the first work to address OOD recommendation via a test-time-training strategy and conducts experiments on five datasets with various backbones.

Abstract

Deep learning has been widely applied in recommender systems, which has achieved revolutionary progress recently. However, most existing learning-based methods assume that the user and item distributions remain unchanged between the training phase and the test phase. However, the distribution of user and item features can naturally shift in real-world scenarios, potentially resulting in a substantial decrease in recommendation performance. This phenomenon can be formulated as an Out-Of-Distribution (OOD) recommendation problem. To address this challenge, we propose a novel Dual Test-Time-Training framework for OOD Recommendation, termed DT3OR. In DT3OR, we incorporate a model adaptation mechanism during the test-time phase to carefully update the recommendation model, allowing the model to specially adapt to the shifting user and item features. To be specific, we propose a self-distillation task and a contrastive task to assist the model learning both the user's invariant interest preferences and the variant user/item characteristics during the test-time phase, thus facilitating a smooth adaptation to the shifting features. Furthermore, we provide theoretical analysis to support the rationale behind our dual test-time training framework. To the best of our knowledge, this paper is the first work to address OOD recommendation via a test-time-training strategy. We conduct experiments on three datasets with various backbones. Comprehensive experimental results have demonstrated the effectiveness of DT3OR compared to other state-of-the-art baselines.

Figures (13)

• Figure 1: Illustration of OOD recommendation. The distribution shift of user and item features in test phase, leading to decline in recommendation performance.
• Figure 2: Illustration of user/item feature shift.
• Figure 3: The overall framework for our proposed DT3OR.
• Figure 4: Illustration of the self-distillation task designing. In our proposed method, we initially obtain the user embedding $\textbf{E}_u$ and item embedding $\textbf{E}_i$ by encoding the user/item features and historical interactions with the training network. After this, we carry out K-means on the combined embedding to yield the clustering results. Subsequently, with the sharpen function, we implement self-distillation to enhance the user interest centers, thus improving the generalization of model to deal with shift features.
• Figure 5: Illustration of the contrastive task designing. With the high confidence results of clustering, we select the samples in the same interest center as positive sample, while regarding the different high confidence interest centers as negative samples.

Theorems & Definitions (5)

• Definition 1: Test-time training for OOD Recommendation (T3OR).
• Theorem 1
• proof
• Theorem 2
• proof