Fernanda Goyo Tamanaka

Through artificial intelligence (AI) techniques, synthetic content is generated, which is an increasingly critical issue for digital security. This paper aims to classify real and synthetic facial images using a state-of-the-art AI model to analyze, both quantitatively and qualitatively, its performance and the most discriminating features across the various data transformations used for this purpose in current literature. In this scenario, the study utilized three public benchmark datasets: two for generating fake images (Stable Diffusion and StyleGAN2) and one for real images (FFHQ), employing the Xception architecture with the Grad-CAM++ viewer. Overall, the results indicate that the central region of the image is the most discriminative one, whereas random cropping produced the worst performance. Future work aims to explore other explainable Artificial Intelligence (XAI) methods, as well as other training and validation protocols, such as cross-testing between synthetic databases.

Newborns feel pain and the more premature they are, the more immature are their pain attenuation system. Facial expression recognition provides a non-invasive method that has been widely used in clinical practice to assess pain in newborns. In this context, this work has performed a qualitative and quantitative assessment between the ranking importance of neonatal facial features proposed by 13 clinical scales commonly cited in the literature and the visual perception of 143 health and non-health professionals on those same features. The degree of similarity between these two ranking approaches was evaluated by Kendall’s correlation. To the clinical scales, the most relevant facial features are the eyes, regions in-between eyebrows and forehead. However, according to the visual perception of the participants when assessing pain on face images, regardless of clinical expertise, the most relevant facial feature is the nose. These qualitative results are statistically supported by the Kendall analyses carried out. Specifically, our results highlight that the neonatal facial features ranked as the most important ones by the clinical scales cannot be considered correlated with the visual perception of the participants, either experts or non-experts. Moreover, all groups of participants have presented high correlated rankings between each other.

Constructo do processo de aprendizagem dos alunos do curso de administração / Construct of the learning process of business students