[PhD Thesis 2021 offer] Quantum Machine Learning Algorithms for Identifying Deepfakes
Supervision: Kamel Smaïli (firstname.lastname@example.org)
Support: competition for a doctoral contract
Google claimed that some problems necessitate 10000 years of calculation (with the fastest computer on the world) but on Google’s quantum computer they necessitate only 200 seconds. These new computers are based on the principle of quantum mechanics [Mora-Lopez 2009]. Quantum computing is modeled by quantum circuits, which are based on the quantum bit, or qubit, which is somewhat analogous to the bit in classical computation. Qubits can be in a 1 or 0 quantum state, or they can be in a superposition of the 1 and 0 states.
Classical computers perform deterministic classical operations or can emulate probabilistic processes using sampling methods. By coupling superposition and entanglement, quantum computers can perform quantum operations that are difficult to emulate at scale with classical computers.
In this Phd thesis, we are interested by developing a new predictive machine learning model based on Quantum Computing. In other words we are interested by quantum machine learning algorithms. Quantum Machine Learning is an emerging interdisciplinary research area, which is the combination of quantum physics and machine learning Quantum Information algorithms are used within the machine learning and artificial intelligence in the state of Quantum system [Shuld 2018]. When Machine learning and Quantum Computer interact with each other, that is called a Quantum Machine Learning [Ciliberto 2018]. Using Quantum Machine Learning (QML) makes a revolution in performance and speed.
Deepfake is a combination of the terms Deep learning and Fake. This term references the audiovisual works in which the image of somebody is synthesized. Actually, deepfakes concern the process of fabrication and manipulation of digital images and videos. by using DNNs that makes the process more convincing. These videos can create illusions of a person’s presence and activities that do not occur [Nguyen 2020], [Yang 2020].
The experiment setup of this Phd will have the goal of detecting deepfake videos. This experiment will test deep and reinforcement learning techniques that use algorithms similar to the ones used to build the deepfake to detect them, such as the GAN method [Goodfellow 2014], [Gauthier 2015]. The metrics will be mainly the accuracy recognizing patterns in howdeepfakes are created, and the number of inconsistencies that the algorithm can pick up.
The PhD student will be under the supervision of Kamel Smaïli, Professor.
The Phd Student will work in the SMarT research group that have a strong experience in deep learning and started recently working on Deepfakes by building the databases necessary. Obviously we do not have a quantum computer, but we will be working on different Quantum computers are online available like IBM Qiskit, Google Quantum AI, Microsoft Azure Quantum Computer, D-Wave and Rigetti Forest.
[Mora-Lopez 2009] J. Mora-Lopez, “Fundamentals of Physics “– Volume I, EOLSS Publications, 2009.
[Schuld 2009] M. Schuld and F. Petruccione, “Supervised learning with quantum computers”, Quantum Science and Technology, Springer, 2018.
[Ciliberto 2018] C. Ciliberto, M. Herbster, A. Alessandro, M. Pontil, A. Rocchetto, S. Severini, L. Wossnig, “Quantum machine learning: a classical perspective”, Proceedings of the Royal Society A: Mathematical, Physical, 2018
[Gauthier 2015] J. Gauthier, “Conditional generative adversarial nets for convolutional face generation”, Technical report, 2015.
[Goodfellow 2014] I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, “Generative adversarial nets”, NIPS, 2014.
[Nguyen 2020] T.T. Nguyen, C.M. Nguyen, D.T. Nguyen, S. Nahavandi, “Deep Learning for Deepfakes Creation and Detection: A Survey”, arXiv, 2020
[Li 2020] Y. Li, X. Yang, P. Sun, H. Qi, S. Lyu, “Celeb-DF: A Large-scale Challenging Dataset for DeepFake Forensics”, arXiv, 2020