Conference Paper
PublishedNeural Ordinary Differential Equations for Continuous-Time Series Analysis in Distributed Sensing
Continuous Dynamic Modeling
Journal / Venue
International Conference on Learning Representations (ICLR)
Paper Link
Not AvailableICLR 2024
Journal Metrics
Metrics Updated: April 2024ICLR Foundation
Keywords
Neural ODEsTime-SeriesIoTDistributed Sensing
Authors
Alex Rivera
hello@alexrivera.meSophia Rodriguez
Overview
Applying Neural ODEs to model irregular time-series data from distributed sensor networks.
Abstract
Neural Ordinary Differential Equations (NODEs) offer a promising approach for modeling continuous-time dynamics. In this work, we extend NODEs to distributed sensing environments, addressing issues of data sparsity and irregular sampling. Our proposed model outperforms traditional recurrent neural networks on multiple real-world IoT datasets.