NumPy implementation of Thierry Denoeux's Evidence-Theoretic Classifier. Basically a generalization of radial basis function (RBF) networks utilizing Dempster-Shafer theory of evidence. Unlike the original RBFNs, can detect novelty, uncertainty and discriminate between the two.
Original paper:
Denoeux, T. (2000). A neural network classifier based on Dempster-Shafer theory. Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on, 30(2), 131-150.