In this paper, we explore new approaches to combining information encoded within the learned representations of auto-encoders. We explore models that are capable of combining the attributes of multiple inputs such that a resynthesised output is trained to fool an adversarial discriminator for real versus synthesised data. Furthermore, we explore the use of such an architecture in the context of semi-supervised learning, where we learn a mixing function whose objective is to produce interpolations of hidden states, or masked combinations of latent representations that are consistent with a conditioned class label. We show quantitative and qualitative evidence that such a formulation is an interesting avenue of research.

In this paper, we explore new approaches to combining information encoded within the learned representations of auto-encoders. We explore models that are capable of combining the attributes of multiple inputs such that a resynthesised output is trained to fool an adversarial discriminator for real versus synthesised data. Furthermore, we explore the use of such an architecture in the context of semi-supervised learning, where we learn a mixing function whose objective is to produce interpolations of hidden states, or masked combinations of latent representations that are consistent with a conditioned class label. We show quantitative and qualitative evidence that such a formulation is an interesting avenue of research.

This paper proposes a method for enhancing the latent space learned by auto-encoders, so that the learned latent space produces meaningful features useful for downstream tasks. The proposed approach considers interpolations in the latent space and encourages the reconstructions from these interpolations to be similar to the data using adversarial learning. The learned latent space is shown to capture useful feature via experiments on MNIST, KMNIST and SVHN. The paper presents some promising preliminary experiments. However, there are many issues in the experimental setup Why is the quality of features measured during training?

The paper explores the following question: If an autoencoder is learned with adversarial training where the inputs to the discriminator is not the reconstruction from autoencoder but that of a reconstruction using interpolations of pairs (or more) of encodings of the training examples, would that lead to better representation learning? Results on simpler datasets showcases efficacy, while at the same time, evaluating the approach on more complex/real-world datasets would make the paper more compelling. The paper can also benefit from rigorour analysis of the Bernoulli mixup. Aside: crossover in biology happens at recombination hotspots and not at random. They are much more structured.

In this paper, we explore new approaches to combining information encoded within the learned representations of auto-encoders. We explore models that are capable of combining the attributes of multiple inputs such that a resynthesised output is trained to fool an adversarial discriminator for real versus synthesised data. Furthermore, we explore the use of such an architecture in the context of semi-supervised learning, where we learn a mixing function whose objective is to produce interpolations of hidden states, or masked combinations of latent representations that are consistent with a conditioned class label. We show quantitative and qualitative evidence that such a formulation is an interesting avenue of research.

In this paper, we explore new approaches to combining information encoded within the learned representations of auto-encoders. We explore models that are capable of combining the attributes of multiple inputs such that a resynthesised output is trained to fool an adversarial discriminator for real versus synthesised data. Furthermore, we explore the use of such an architecture in the context of semi-supervised learning, where we learn a mixing function whose objective is to produce interpolations of hidden states, or masked combinations of latent representations that are consistent with a conditioned class label. We show quantitative and qualitative evidence that such a formulation is an interesting avenue of research. Papers published at the Neural Information Processing Systems Conference.