Branched Variational Autoencoder Classifiers
This work addresses classification accuracy enhancement in machine learning, particularly for image datasets like MNIST, but it is incremental as it modifies existing VAE architectures.
The paper tackles the problem of improving classification accuracy by introducing a branched variational autoencoder (BVAE) that incorporates class labels into the loss function, resulting in separated and ordered latent space distributions; numerical calculations on the MNIST dataset show improved performance, including for rotated digits and with various output distributions.
This paper introduces a modified variational autoencoder (VAEs) that contains an additional neural network branch. The resulting branched VAE (BVAE) contributes a classification component based on the class labels to the total loss and therefore imparts categorical information to the latent representation. As a result, the latent space distributions of the input classes are separated and ordered, thereby enhancing the classification accuracy. The degree of improvement is quantified by numerical calculations employing the benchmark MNIST dataset for both unrotated and rotated digits. The proposed technique is then compared to and then incorporated into a VAE with fixed output distributions. This procedure is found to yield improved performance for a wide range of output distributions.