Meta-Learning with Adaptive Weighted Loss for Imbalanced Cold-Start Recommendation

By Minchang Kim et al

Published on Oct. 21, 2025

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Related Work
3. Preliminary
4. Methodology
4.1 Meta-Learner
4.2 Adaptive Weighted Loss
5. Conclusion

Summary

The paper discusses the challenges of cold-start recommendations in sequential recommenders and proposes a novel framework, MELO, that addresses imbalanced rating distributions. It introduces a meta-learner based on MAML for fast adaptation to new users with limited interactions. Additionally, an adaptive weighted loss mechanism is presented to correct the learning process based on imbalanced data. Experimental validation on real-world datasets demonstrates the effectiveness of the framework.