Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

O2, repositorio institucional de la UOC
Publicación Contribución de congreso (ConferenceOutput)

Toward universal detector for synthesized images by estimating generative AI models

  • Seo, Ryota
  • Kuribayashi, Minoru
  • Ura, Akinobu
  • Mallet, Antoine
  • Cogranne, Rémi
  • Mazurczyk, Wojciech
  • Megias, David
One of the vulnerabilities in discriminators for AI-generated images is that the classification accuracy degrades when dealing with images generated using methods other than those they were trained on. As a countermeasure, in this study, we propose an image generation method estimator. The process of discrimination involves the input of an image to the estimator, which estimates the method used for its generation. Subsequently, a specialized fake image discriminator tailored to the estimated image generation method is used to identify the authenticity of the image. The activation functions are also considered according to the estimation results and analyzed for those discriminators. Discrimination scores are weighted and aggregated according to the estimation results, and the final decision is output. Our experimental results showed that the estimator achieved a classification accuracy of approximately 90% for 18 types of AI-generated images. Furthermore, by selecting the top two estimations in order of confidence, the accuracy increased to around 98%.
O2, repositorio institucional de la UOC
Publicación Artículo científico (JournalArticle)

Enhancing unsupervised anomaly-based cyberattacks detection in smart homes through hyperparameter optimization

  • Iturbe Araya, Juan Ignacio
  • Rifà-Pous, Helena
As smart homes become increasingly interconnected, it is crucial to ensure their security against cyber threats. This research focuses on improving anomaly detection within smart home environments through the application of unsupervised learning methods. Unlike traditional methods that require prior knowledge of specific attack types, unsupervised methods learn from normal operational data, identifying anomalies as deviations from this norm. This study investigates the impact of hyperparameter tuning on the effectiveness of these methods compared to their default settings. This research aims to identify optimal strategies for the configuration of unsupervised learning algorithms in smart home environments. We applied four models (Elliptic Envelope, Isolation Forest, Local Outlier Factor, and One-class SVM) in four datasets (Bot-IoT, IoTID20, N-baiot, and Ton-IoT). Our findings indicate a significant improvement in performance with hyperparameter optimization.