Deepfake Detection of Face Images based on a Convolutional Neural Network

Vocabulary flashcards related to deepfake detection using convolutional neural networks, transfer learning, and relevant datasets.

Deepfake

Falsification of content, usually images and videos, using generative machine learning algorithms.

ResNet-50

A residual convolutional neural network model consisting of 50 parameter adoptable convolutional layers paired with residual connections.

Transfer Learning

A machine learning technique where a model developed for a first task is reused as the starting point for a model on a second task.

Diverse Face Fake Dataset (DFFD)

A dataset consisting of real and fake portrait images of human faces, containing diversity in terms of age, face size, race, and manipulation types.

Binary Classification

A classification task with two possible outcomes, often used to distinguish between authentic and fake images.

Area Under Curve (AUC)

A metric used to evaluate the performance of a binary classification model, representing the area under the Receiver Operating Characteristic (ROC) curve.

Convolutional Neural Network (CNN)

A type of neural network commonly used for image recognition and processing, utilizing convolutional layers to extract features from images.

Fine Tuning

A process where a pre-trained model's weights are adjusted on a new dataset.

Sigmoid Activation Function

Takes the input and returns a value between 0 and 1. Used for binary classification task.

Adam Optimizer

An optimization algorithm that adapts the learning rate during the training process, making it an effective optimization method for deep learning models.

Binary Cross Entropy (Log Loss)

Loss function that is commonly used for measuring the performance of binary classification problems.

Kernel Density Estimation (KDE)

Estimates the probability density function with the use of the guassian kernel to get an estimation close to the real PDF.