MSc Thesis Proposal Announcement of Ehsan Ur Rahman Mohammed:"Face Image Generation using Attention and Knowledge Transfer "

Monday, November 14, 2022 - 10:00 to 11:30

SCHOOL OF COMPUTER SCIENCE 

The School of Computer Science is pleased to present… 

MSc Thesis Proposal by: Ehsan Ur Rahman Mohammed  

 
Date: Monday November 14, 2022 
Time:  10:00AM – 11:30AM 
Location: Essex Hall, Room 105 
Reminders: Two-part attendance mandatory, arrive 5-10 minutes prior to event starting - LATECOMERS WILL NOT BE ADMITTED once the door has been closed and the presentation has begun. Please be respectful of the presenter by NOT knocking on the closed door for admittance.
 

Abstract:  

The success of deep learning relies on huge datasets and access to immense computational power. However, in some domains, data is not always accessible due to privacy concerns or lack of a mechanism to collect and store data safely, among many other constraints. To alleviate this problem of data scarcity, data generation has evolved. One such task that faces data scarcity is image generation, specifically face image generation, due to issues such as privacy. Face image generation has many applications, such as creating huge datasets for applications like image super-resolution, image completion, and image matting. Real-world application of face image generation includes video surveillance, biometrics, and in medical applications such as mapping facial features to genetic data. Generative adversarial networks (GANs) have been used to generate images in the recent literature. In this work, GANs are used for the task of face image generation. GANs have two networks, a discriminator network, and a generator network. The discriminator distinguishes real data, i.e., data from the dataset, from fake data, i.e., data coming from the generator. The generator generates synthetic data. The proposed work aims to utilize attention and knowledge transfer to generate images of faces resembling images from the CelebA dataset. The unique point of the proposed solution is the concept of using pre-trained models as initial models for generating images in domains that lack huge amounts of data. The proposed work will use an autoencoder-based model previously trained on a related dataset. The trained layers from this pre-trained model will be used in both the discriminator and the generator to eliminate the need for training from scratch. The ramifications are reduced training time, computational burden, and the ability to reuse deep learning models. In addition to knowledge transfer as described above, self-attention is used in the GAN model to improve the quality of generated images. Self-attention drastically reduces the number of model parameters that need to be learnt separately. It also helps learn filters with more focus on the important parts of the image, which is especially important for facial image generation. As face images have a distinct foreground and background, the impact of using self-attention becomes more profound.   
 
Keywords: transfer learning, generative adversarial networks, attention, image generation  


MSc Thesis Committee:  

Internal Reader: Dr. Alioune Ngom 
External Reader: Dr. Abdulkudir A. Hussein          
Advisor: Dr. Imran Ahmad 


 MSc Thesis Proposal Announcement 

 

5113 Lambton Tower 401 Sunset Ave. Windsor ON, N9B 3P4 (519) 253-3000 Ext. 3716 csgradinfo@uwindsor.ca