Epistemic Uncertainty-aware Recommendation Systems via Bayesian Deep Ensemble Learning
This addresses recommendation system limitations for users and platforms by providing more robust predictions, though it appears to be an incremental hybrid approach combining existing techniques.
The paper tackles the problems of overfitting and lack of epistemic uncertainty in recommendation systems by proposing BDECF, a Bayesian Deep Ensemble Collaborative Filtering method that combines Bayesian Neural Networks, attention-based matching, and ensemble learning, achieving improved performance across multiple real-world datasets with varying sparsity.
Recommending items to users has long been a fundamental task, and studies have tried to improve it ever since. Most well-known models commonly employ representation learning to map users and items into a unified embedding space for matching assessment. These approaches have primary limitations, especially when dealing with explicit feedback and sparse data contexts. Two primary limitations are their proneness to overfitting and failure to incorporate epistemic uncertainty in predictions. To address these problems, we propose a novel Bayesian Deep Ensemble Collaborative Filtering method named BDECF. To improve model generalization and quality, we utilize Bayesian Neural Networks, which incorporate uncertainty within their weight parameters. In addition, we introduce a new interpretable non-linear matching approach for the user and item embeddings, leveraging the advantages of the attention mechanism. Furthermore, we endorse the implementation of an ensemble-based supermodel to generate more robust and reliable predictions, resulting in a more complete model. Empirical evaluation through extensive experiments and ablation studies across a range of publicly accessible real-world datasets with differing sparsity characteristics confirms our proposed method's effectiveness and the importance of its components.