This project presents an improved algorithm, named IB-RVFL, which is a complete Bayesian framework combined with the Random Vector Functional-Link (RVFL) networks for uncertain data modeling. In comparison to the existing work, we address the use of prior distributions on the parameters of basis functions. This additional information helps to enhance the learning power of the model and provides an effective solution for the difficult and significant setting problem of random parameters in existing RVFL-based modelling techniques. A Variational Inference (VI) method is used to obtain an approximation of the intractable posterior distribution, which helps to realize automatic inference of the hyper-parameters and gives a probability estimate for the test data.
In our experiments, we first consider a comprehensive comparison with RVFL, B-RVFL and IB-RVFL networks based on five real-world regression data sets used in "Bayesian random vector functional-link networks for robust data modeling". Then we add four new data sets to compare IB-RVFL, B-RVFL, RSCNs in "Robust stochastic configuration networks with kernel density estimation for uncertain data regression" and other two classical machine learning algorithms-SVM, Random Forest. All the data sets come from UCI{http://archive.ics.uci.edu/ml/datasets.html} and KEEL{http://sci2s.ugr.es/keel/datasets.php}.
If you want to run the IB-RVFL algorithm, please first install the library pymc3. We give some examples in the file "example1.py" and "example2.py". And you can find the B-RVFL algorithm in B-RVFL.py, the RSCNs algorithm in SCN-I.py.