Biblio

Style transfer is an emerging trend in the fields of deep learning's applications, especially in images and audio data this is proven very useful and sometimes the results are astonishing. Gradually styles of textual data are also being changed in many novel works. This paper focuses on the transfer of the sentimental vibe of a sentence. Given a positive clause, the negative version of that clause or sentence is generated keeping the context same. The opposite is also done with negative sentences. Previously this was a very tough job because the go-to techniques for such tasks such as Recurrent Neural Networks (RNNs) [1] and Long Short-Term Memories(LSTMs) [2] can't perform well with it. But since newer technologies like Generative Adversarial Network(GAN) and Variational AutoEncoder(VAE) are emerging, this work seem to become more and more possible and effective. In this paper, Multi-Genarative Variational Auto-Encoder is employed to transfer sentiment values. Inspite of working with a small dataset, this model proves to be promising.

We explore methods of producing adversarial examples on deep generative models such as the variational autoencoder (VAE) and the VAE-GAN. Deep learning architectures are known to be vulnerable to adversarial examples, but previous work has focused on the application of adversarial examples to classification tasks. Deep generative models have recently become popular due to their ability to model input data distributions and generate realistic examples from those distributions. We present three classes of attacks on the VAE and VAE-GAN architectures and demonstrate them against networks trained on MNIST, SVHN and CelebA. Our first attack leverages classification-based adversaries by attaching a classifier to the trained encoder of the target generative model, which can then be used to indirectly manipulate the latent representation. Our second attack directly uses the VAE loss function to generate a target reconstruction image from the adversarial example. Our third attack moves beyond relying on classification or the standard loss for the gradient and directly optimizes against differences in source and target latent representations. We also motivate why an attacker might be interested in deploying such techniques against a target generative network.