Evaluation of Regularization Techniques in Generative Adversarial Networks

Student: Valchuk Kseniia

Supervisor: Alexey Umnov

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Year of Graduation: 2018

The key problem of Generative Adversarial Networks (GANs) is the instability during the training procedure. As a result, generated samples are unrealistic or generative adversarial networks do not converge. In order to improve the stability of these models, we will try to apply such methods as Binary Dropout and Variational Dropout to Deep Convolutional Generative Adversarial Networks. The primary goal of this work is to compare regularized generative models with models without regularization and to explore the difference between these training procedures to reveal whether proposed regularization techniques will be helpful or not.

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses