EvtGraph: Event-Adaptive Compression for Sparse Temporal Graph Learning in Multimodal Time Series

2026-08-04

AI Summary

This work addresses the pervasive challenges of uneven information density and redundancy in multimodal time-series data, which render conventional uniform discretization methods inefficient. To overcome these limitations, the authors propose EvtGraph, a novel framework that introduces, for the first time, an event-centric representation paradigm under explicit computational budget constraints. The approach employs Event-Adaptive Modal Compression (EAMC) to reparameterize raw time-series into event-level tokens, selects a critical subset via Node Budget Control (NBC), and leverages Temporal-constrained Sparse Graph (T2SG) reasoning for efficient modeling. Evaluated on multimodal clinical benchmarks such as MIMIC-IV+CXR and cross-domain datasets, EvtGraph significantly outperforms Transformer- and recurrent network-based baselines, achieving higher performance with substantially lower computational overhead.

Abstract

Multimodal temporal data are inherently irregular and uneven in information density, yet most models rely on uniform discretization, leading to inefficient representations. We propose \textbf{EvtGraph}, a unified framework that aligns computation with temporal salience under explicit budget constraints. EvtGraph reparameterizes sequences into event-level tokens via event-adaptive compression (EAMC), selects a compact subset with a node budget (NBC), and performs temporally constrained sparse graph reasoning (T2SG). This transforms dense sequences into structured computation over salient events, reducing complexity while preserving critical transitions. We show that this design provides a practical mechanism for allocating representational capacity under a fixed budget, yielding a consistent performance--efficiency trade-off, where a small budget is often sufficient in practice. Experiments on multimodal clinical (MIMIC-IV + CXR) and cross-domain benchmarks demonstrate that EvtGraph outperforms both Transformer-based and recurrent baselines while significantly improving efficiency. These results suggest that budget-constrained event-centric representation provides a general paradigm for learning from high-redundancy temporal data.

Problem

• multimodal time series
• irregular temporal data
• information density
• uniform discretization
• representation efficiency

Innovation

• event-adaptive compression
• sparse temporal graph
• budget-constrained learning
• multimodal time series
• temporal salience