Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Markov Processes for Enhanced Deepfake Generation and Detection

Michael A. Kouritzin, Ian Zhang, Jyoti Bhadana, Seoyeon Park

Abstract

New and existing methods for generating, and especially detecting, deepfakes are investigated and compared on the simple problem of authenticating coin flip data. Importantly, an alternative approach to deepfake generation and detection, which uses a Markov Observation Model (MOM) is introduced and compared on detection ability to the traditional Generative Adversarial Network (GAN) approach as well as Support Vector Machine (SVM), Branching Particle Filtering (BPF) and human alternatives. MOM was also compared on generative and discrimination ability to GAN, filtering and humans (as SVM does not have generative ability). Humans are shown to perform the worst, followed in order by GAN, SVM, BPF and MOM, which was the best at the detection of deepfakes. Unsurprisingly, the order was maintained on the generation problem with removal of SVM as it does not have generation ability.

Markov Processes for Enhanced Deepfake Generation and Detection

Abstract

New and existing methods for generating, and especially detecting, deepfakes are investigated and compared on the simple problem of authenticating coin flip data. Importantly, an alternative approach to deepfake generation and detection, which uses a Markov Observation Model (MOM) is introduced and compared on detection ability to the traditional Generative Adversarial Network (GAN) approach as well as Support Vector Machine (SVM), Branching Particle Filtering (BPF) and human alternatives. MOM was also compared on generative and discrimination ability to GAN, filtering and humans (as SVM does not have generative ability). Humans are shown to perform the worst, followed in order by GAN, SVM, BPF and MOM, which was the best at the detection of deepfakes. Unsurprisingly, the order was maintained on the generation problem with removal of SVM as it does not have generation ability.

Paper Structure

This paper contains 29 sections, 23 equations, 4 figures, 3 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Layout
  3. Computer System
  4. Methods of Data Generation
  5. Initial Data
  6. Real Sequences
  7. Real Fakes
  8. Initial Deepfakes, the Filtering Generator
  9. One Step Learning
  10. Generative Learning with GAN
  11. MOM Learning
  12. GAN's Generator
  13. MOM's Generator
  14. Methods of Classifying a coin-flip
  15. GAN Discriminator
  16. ...and 14 more sections

Figures (4)

  • Figure 1: Workflow of the proposed MOM-based framework on a binary-sequence example ($N=200$). The method consists of three stages: (i) MOM parameter learning via EM (forward/backward recursions and updates of $p,q,\mu$), (ii) sequence generation using fitted MOM parameters, and (iii) classification by recursive-likelihood (Bayes-factor) scoring against class-specific model banks, followed by label assignment.
  • Figure 2: Generation in MOM
  • Figure 3: Discriminator in MOM model
  • Figure 4: Distribution of the results of the coin flip quiz