pg-presentation-template.txt

File contents

Template for Presentation Details 

Below is an example of the information to be sent to the Postgraduate Coordinator and the Postgraduate Teaching Assistant:


MSc Proposal Presentation

Monday, 26 February 2018 | 3pm
Gary Marsden Boardroom, ICT4D Centre

Supervisor: Prof. Deshen Moodley

ABSTRACT:
Generative Adversarial Networks (GAN) are a generative modelling technique used in many image-related domains. Recently, GAN have been used for fine art generation. Modern GAN variants have been used to generate art of a particular genre, artist or style. These GAN are evaluated quantitatively using common GAN evaluation metrics. Elgammal et al.'s Creative Adversarial Network (CAN), the foremost fine art GAN, has been used to generate style-ambiguous art and is evaluated qualitatively, using MTurk surveys. This thesis aims to apply modern GAN variants to create style-ambiguous, style-distorted and style-specific art, and to evaluate the result from three perspectives: whether the GAN's images are indeed art, the quality of the GAN by common GAN metrics, and the creativity of the GAN, using accepted evaluation metrics of computational creativity.

The email you need to circulate should look like this:

The subject line: Alan Berman - MSc Proposal Presentation:  26 February | 3pm

The body of the email:

MSc Proposal Presentation

Monday, 26 February 2018 | 3pm
Gary Marsden Boardroom, ICT4D Centre

Supervisor: Prof. Deshen Moodley

ABSTRACT:
Generative Adversarial Networks (GAN) are a generative modelling technique used in many image-related domains. Recently, GAN have been used for fine art generation. Modern GAN variants have been used to generate art of a particular genre, artist or style. These GAN are evaluated quantitatively using common GAN evaluation metrics. Elgammal et al.'s Creative Adversarial Network (CAN), the foremost fine art GAN, has been used to generate style-ambiguous art and is evaluated qualitatively, using MTurk surveys. This thesis aims to apply modern GAN variants to create style-ambiguous, style-distorted and style-specific art, and to evaluate the result from three perspectives: whether the GAN's images are indeed art, the quality of the GAN by common GAN metrics, and the creativity of the GAN, using accepted evaluation metrics of computational creativity.