Evidence Theory

Evidence theory, encompassing Dempster-Shafer theory and belief functions, offers a framework for representing and managing uncertainty, particularly when dealing with incomplete or conflicting information. Current research focuses on improving the accuracy and efficiency of evidence fusion methods, including the development of novel architectures like the Time Evidence Fusion Network and the application of evidence theory to enhance existing models such as ensemble classifiers and Bayesian networks. This work is impacting diverse fields, from automated driving (improving occupancy map prediction) to time series forecasting and anomaly detection, by providing more robust and interpretable results with associated uncertainty quantification.