Pretext Training Algorithms for Event Sequence Data
This addresses the need for effective representation learning in event sequence analysis, though it appears incremental by adapting existing pretext training approaches to a specific data type.
The paper tackles the problem of learning representations for event sequence data by proposing a self-supervised pretext training framework with a novel alignment verification task, achieving competitive performance across tasks like next-event prediction and classification on public benchmarks.
Pretext training followed by task-specific fine-tuning has been a successful approach in vision and language domains. This paper proposes a self-supervised pretext training framework tailored to event sequence data. We introduce a novel alignment verification task that is specialized to event sequences, building on good practices in masked reconstruction and contrastive learning. Our pretext tasks unlock foundational representations that are generalizable across different down-stream tasks, including next-event prediction for temporal point process models, event sequence classification, and missing event interpolation. Experiments on popular public benchmarks demonstrate the potential of the proposed method across different tasks and data domains.