DynamiX: Dynamic Resource eXploration for Personalized Ad-Recommendations

2025-11-23

AI Summary

In online advertising recommendation, modeling full user behavior sequences incurs high computational overhead and is sensitive to noise. To address this, we propose DynamiX—a scalable personalized sequential modeling framework that pioneers the joint integration of session-level and interface-level modeling. DynamiX introduces an event-level feature pruning and enhancement mechanism guided by dwell time and conversion correlation, along with event-based features (EBFs), self-supervised user clustering, dynamic feature boosting, and resource removal techniques to achieve co-optimization of training and inference. Experiments demonstrate that DynamiX improves training and inference throughput by 1.15% and 1.8%, respectively, increases inference QPS by 4.2%, and reduces normalized entropy (NE) by 0.033—achieving substantial computational cost reduction while preserving prediction accuracy.

Technology Category

• Machine Learning: Online Learning & Bandits
• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

For online ad-recommendation systems, processing complete user-ad-engagement histories is both computationally intensive and noise-prone. We introduce Dynamix, a scalable, personalized sequence exploration framework that optimizes event history processing using maximum relevance principles and self-supervised learning through Event Based Features (EBFs). Dynamix categorizes users-engagements at session and surface-levels by leveraging correlations between dwell-times and ad-conversion events. This enables targeted, event-level feature removal and selective feature boosting for certain user-segments, thereby yielding training and inference efficiency wins without sacrificing engaging ad-prediction accuracy. While, dynamic resource removal increases training and inference throughput by 1.15% and 1.8%, respectively, dynamic feature boosting provides 0.033 NE gains while boosting inference QPS by 4.2% over baseline models. These results demonstrate that Dynamix achieves significant cost efficiency and performance improvements in online user-sequence based recommendation models. Self-supervised user-segmentation and resource exploration can further boost complex feature selection strategies while optimizing for workflow and compute resources.

Problem

• Optimizes ad-recommendation by processing user engagement histories efficiently
• Reduces computational intensity and noise in online recommendation systems
• Enhances training and inference efficiency while maintaining prediction accuracy

Innovation

• Dynamic resource exploration optimizes event history processing
• Self-supervised learning categorizes user engagements at multiple levels
• Selective feature removal and boosting improves efficiency and accuracy