ASAP, an interactive visual analytics system, enables efficient identification and in-depth analysis of distinctive patterns in AI-generated images to support the discernment of authentic from synthetic content.
WaveMix is a novel neural architecture that achieves comparable or better accuracy than state-of-the-art convolutional neural networks, vision transformers, and token mixers for various vision tasks, while using fewer trainable parameters, GPU RAM, and computations.
Explainable AI is used to red team hyperspectral image analysis models, identifying flaws and biases for improved performance.
Proposing a framework to determine the likelihood of an image being fake based on changes in image metadata.