A Novel Variational Autoencoder with Applications to Generative Modelling, Classification, and Ordinal Regression
This work addresses generative and classification problems in machine learning, but it appears incremental as it builds on existing variational autoencoder approaches with specific modifications.
The authors developed a novel variational autoencoder model with a new latent prior specification and an ordinality enforcing unit, tackling generative modeling, classification, and ordinal regression. Their results show comparable performance to baselines on two benchmark datasets for classification tasks.
We develop a novel probabilistic generative model based on the variational autoencoder approach. Notable aspects of our architecture are: a novel way of specifying the latent variables prior, and the introduction of an ordinality enforcing unit. We describe how to do supervised, unsupervised and semi-supervised learning, and nominal and ordinal classification, with the model. We analyze generative properties of the approach, and the classification effectiveness under nominal and ordinal classification, using two benchmark datasets. Our results show that our model can achieve comparable results with relevant baselines in both of the classification tasks.