Implementantion of the neural network architecture described in the paper (doi.org/10.48550/arXiv.2310.08055). The network uses a Bayesian convolutional layer combined with deterministic convolutional and dense layers and a probabilistic output layer. The output layer is modelled as a gamma process. For more details, see the paper above and the Specnet architecture and the associated paper.
For generating the training data sets, see this repository.
If you find the software useful, please cite (doi.org/10.48550/arXiv.2310.08055).