College of Computer Science and Mathematics

Defense of a PhD thesis in the College of Computer Science and Mathematics – Department of Computer Science

Deepfake Detection based on Machine Learning

(Deepfake detection using machine learning)

In continuation of the scientific research movement and the follow-up and presence of the Dean of the College of Computer Science and Mathematics, Prof. Dr. Duha Bashir Abdullah

Discussed in the discussion hall of the College of Computer Science and Mathematics at the University of Mosul on Thursday 11- 7 – 2024

Doctoral thesis (Deepfake Detection based on Machine Learning) by the student Duha Amer Sultan and under the supervision of Prof. Dr. Lahib Muhammad Ibrahim

The thesis submitted by the student dealt with the detection of fake videos generated using artificial intelligence techniques

The study touched on taking advantage of the audio and visual features of the video. Two deepfakes detection models were built:

1- Unimodal: which works in two ways: First: It relies on audio features only from the video and extracted using the MFCC function. Second: It depends on visual features only for faces cut from video frames, and two methods have been adopted to extract visual features: FaceNet + PCA+ HeadPose estimation

VGGFace+PCA+HeadPose estimation

2- Multimodal: The previous two models were combined to obtain a multi-model which greatly improved the final accuracy of the work

LazyClassifier relied on distinguishing the video whether it was real or fake

The study aims to. Reduce the risk of deepfakes in phishing, impersonation and the dissemination of disinformation

 

The discussion committee was chaired by Prof. Dr. Fawzia Mahmoud Ramo

Prof. Dr.  Ibrahim Ahmed Saleh Member

Prof. Dr.  Student Charter Kata Member

Assoc. Prof. Jamal Salah El-Din Majeed Member

Assoc. Prof. Elaf Osama Abdel Majeed Member

Prof. Lahib Mohamed Ibrahim Member and Supervisor