Accurate Bayesian Data Classification without Hyperparameter Cross-validation
This work addresses the computational burden of hyperparameter tuning in Bayesian classification, which is a problem for practitioners in machine learning, though it appears incremental as it builds on existing Bayesian discriminant analysis methods.
The authors tackled the problem of hyperparameter tuning in Bayesian data classification by extending the standard model with a generalized prior and deriving hyperparameters analytically via evidence maximization, achieving competitive accuracy with state-of-the-art methods while eliminating the computational cost of cross-validation.
We extend the standard Bayesian multivariate Gaussian generative data classifier by considering a generalization of the conjugate, normal-Wishart prior distribution and by deriving the hyperparameters analytically via evidence maximization. The behaviour of the optimal hyperparameters is explored in the high-dimensional data regime. The classification accuracy of the resulting generalized model is competitive with state-of-the art Bayesian discriminant analysis methods, but without the usual computational burden of cross-validation.