Features Fusion Framework for Multimodal Irregular Time-series Events
This work addresses a domain-specific problem in healthcare or sensor data analysis by improving prediction accuracy for irregular time-series events, though it appears incremental as it builds on LSTM with feature fusion techniques.
The paper tackles the challenge of modeling multimodal irregular time-series events with varying frequencies and complex relationships by proposing a features fusion framework based on LSTM, which significantly outperforms existing methods on the MIMIC-III dataset in terms of AUC and AP.
Some data from multiple sources can be modeled as multimodal time-series events which have different sampling frequencies, data compositions, temporal relations and characteristics. Different types of events have complex nonlinear relationships, and the time of each event is irregular. Neither the classical Recurrent Neural Network (RNN) model nor the current state-of-the-art Transformer model can deal with these features well. In this paper, a features fusion framework for multimodal irregular time-series events is proposed based on the Long Short-Term Memory networks (LSTM). Firstly, the complex features are extracted according to the irregular patterns of different events. Secondly, the nonlinear correlation and complex temporal dependencies relationship between complex features are captured and fused into a tensor. Finally, a feature gate are used to control the access frequency of different tensors. Extensive experiments on MIMIC-III dataset demonstrate that the proposed framework significantly outperforms to the existing methods in terms of AUC (the area under Receiver Operating Characteristic curve) and AP (Average Precision).