Addressing Mark Imbalance in Integration-free Neural Marked Temporal Point Processes

2025-10-23

AI Summary

To address the severe label imbalance in mark distributions and poor predictive performance for rare marks in marked temporal point processes (MTPPs), this paper proposes a novel “mark-then-time” paradigm. The core innovation is a learnable thresholding mechanism grounded in mark prior probabilities, enabling efficient modeling and inference for rare events without numerical integration. Integrated within a neural MTPP framework, the method unifies mark selection and time prediction through prior-based normalization, adaptive thresholding, and joint sampling. Extensive experiments on multiple real-world datasets demonstrate significant improvements in both rare-mark classification accuracy and overall temporal prediction precision. Notably, this approach is the first to effectively mitigate label imbalance in MTPPs while maintaining computational efficiency—achieving robust performance without sacrificing scalability or inference speed.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Planning with Markov Models (MDPs, POMDPs)
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

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

Abstract

Marked Temporal Point Process (MTPP) has been well studied to model the event distribution in marked event streams, which can be used to predict the mark and arrival time of the next event. However, existing studies overlook that the distribution of event marks is highly imbalanced in many real-world applications, with some marks being frequent but others rare. The imbalance poses a significant challenge to the performance of the next event prediction, especially for events of rare marks. To address this issue, we propose a thresholding method, which learns thresholds to tune the mark probability normalized by the mark's prior probability to optimize mark prediction, rather than predicting the mark directly based on the mark probability as in existing studies. In conjunction with this method, we predict the mark first and then the time. In particular, we develop a novel neural MTPP model to support effective time sampling and estimation of mark probability without computationally expensive numerical improper integration. Extensive experiments on real-world datasets demonstrate the superior performance of our solution against various baselines for the next event mark and time prediction. The code is available at https://github.com/undes1red/IFNMTPP.

Problem

• Addressing event mark imbalance in temporal point processes
• Improving prediction accuracy for rare event marks
• Developing integration-free neural model for efficient computation

Innovation

• Thresholding method tunes mark probability using learned thresholds
• Predicts mark first then time in a sequential approach
• Neural MTPP model avoids expensive numerical improper integration