Classification by Re-generation: Towards Classification Based on Variational Inference
This addresses the need for more interpretable and scalable classification methods in machine learning, though it appears incremental as it builds on existing VAE frameworks.
The paper tackles the problem of semantic interpretability and scalability in deep neural networks for classification by proposing a classification by re-generation approach using variational autoencoders, where separate encoder-decoder pairs are trained per class, enabling incremental addition of classes without retraining the entire network and introducing a KL divergence-based rejection criterion to improve trust and handle adversarial examples.
As Deep Neural Networks (DNNs) are considered the state-of-the-art in many classification tasks, the question of their semantic generalizations has been raised. To address semantic interpretability of learned features, we introduce a novel idea of classification by re-generation based on variational autoencoder (VAE) in which a separate encoder-decoder pair of VAE is trained for each class. Moreover, the proposed architecture overcomes the scalability issue in current DNN networks as there is no need to re-train the whole network with the addition of new classes and it can be done for each class separately. We also introduce a criterion based on Kullback-Leibler divergence to reject doubtful examples. This rejection criterion should improve the trust in the obtained results and can be further exploited to reject adversarial examples.