Tag
This paper presents a foundation transformer model pretrained on multimodal event sequences for financial applications, unifying heterogeneous data sources and using next-event prediction to learn general-purpose representations. The approach outperforms traditional task-specific models and was deployed at a major Eastern European bank, yielding measurable business improvements.
This paper introduces SigTPP, the first signature-based generative model for temporal point processes, which uses a pathwise framework and interarrival embedding to overcome limitations of likelihood-based methods, and demonstrates strong empirical performance across synthetic and real-world datasets.
This paper introduces Monotone Alternating Splines (MAS), a novel framework for modeling cumulative conditional intensity functions in temporal point processes. MAS overcomes structural limitations of existing monotone neural networks, achieving better accuracy and efficiency on synthetic and real datasets.
This paper studies scaling laws for behavioral foundation models trained on sequences of user actions, finding that a small event embedder is compute-optimal and that the evaluation metric itself influences the optimal compute allocation.