Deep generative LDA
This work addresses the problem of improving classification and dimension reduction for complex data distributions, but it appears incremental as it builds on a previously proposed DNF model.
The authors tackled the limitations of linear discriminant analysis (LDA) in handling complex data distributions by reinterpreting a discriminative normalization flow (DNF) model as a deep generative LDA, showing it is much more powerful than conventional LDA in modeling complex data and retrieving low-dimensional representations.
Linear discriminant analysis (LDA) is a popular tool for classification and dimension reduction. Limited by its linear form and the underlying Gaussian assumption, however, LDA is not applicable in situations where the data distribution is complex. Recently, we proposed a discriminative normalization flow (DNF) model. In this study, we reinterpret DNF as a deep generative LDA model, and study its properties in representing complex data. We conducted a simulation experiment and a speaker recognition experiment. The results show that DNF and its subspace version are much more powerful than the conventional LDA in modeling complex data and retrieving low-dimensional representations.