解决短视频观看时长预测中的分布不稳定问题，提升推荐效果。
Hierarchical Exponential-Gaussian Mixtures for Watch-Time Distribution Prediction
- 构建分层跳跃观看结构，缓解成分退化与方差坍缩
- 在多个数据集上提升排序准确率与事件预测能力
- 适合关注推荐系统中时长分布建模的工程师与研究者
精准的观看时长（WT）预测对短视频推荐至关重要。然而，观看时长分布具有近零膨胀、长尾和多峰特性。近期的指数-高斯混合网络（EGMN）通过建模完整条件分布而非单一点估计，达到了顶尖性能。我们的大规模复现研究发现，EGMN存在方差坍缩、成分冗余和无效成分等问题。为此，我们提出分层指数-高斯混合（HEGM）模型，通过分层跳过观看分解、基于KL的方差正则化、结构化初始化，移除强制高斯偏移和熵正则项来解决上述缺陷。在公开与大规模工业数据集上，HEGM显著提升排序准确率与阈值事件预测表现，保持竞争力的点估计精度，并大幅增强混合模型的稳定性和可解释性。为期1.5个月的生产环境A/B测试验证了用户参与度的显著提升。代码与模型已开源：https://github.com/rw404/HEGM。
原文摘要 · Abstract (English)
Accurate watch-time (WT) prediction is an important requirement for short-video recommendations. Yet WT distributions are near-zero-inflated, long-tailed and multimodal. The recent Exponential-Gaussian Mixture Network (EGMN) models the full conditional WT distribution rather than a single point estimate and achieves state-of-the-art performance. Our large-scale reproduction study reveals that EGMN is vulnerable to variance collapse, component redundancy, and inactive components. We propose a Hierarchical Exponential-Gaussian Mixture (HEGM) model that addresses these failure modes through a hierarchical skip-watch decomposition, KL-based variance regularization, structured initialization, removing the forced Gaussian shift and the entropy regularizer. Across public and large-scale industrial datasets, HEGM improves ranking accuracy and threshold-event prediction, while maintaining competitive point-estimation accuracy and substantially improving mixture stability and interpretability. A 1.5-month production A/B test confirms statistically significant engagement lifts. Our code and models are publicly released at https://github.com/rw404/HEGM.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。