Explore a cutting-edge approach to time series forecasting in this 54-minute conference talk by Yulia Gel. Delve into the innovative coupling of time-aware multipersistence knowledge representation with Graph Convolutional Networks (GCNs). Learn how this method addresses limitations in existing Graph Neural Networks (GNNs) by explicitly accounting for time dependencies and inferring latent time-conditioned relations among entities. Discover the power of multipersistence in topological data analysis for capturing hidden time-conditioned properties. Understand the construction of a supragraph convolution module that simultaneously considers intra- and inter-dependencies in data. Examine practical applications of this approach in forecasting highway traffic flow, blockchain Ethereum token prices, and COVID-19 hospitalizations. Gain insights into the stability of the time-aware multipersistence Euler-Poincaré surface and its role in enhancing forecasting capabilities.