Speech Title: Image Processing Applications of Generative Models with their Security Research
Abstract:
Image generative models, such as Generative Adversarial Network (GAN) and Variational autoencoder (VAE), have profoundly changed the field of image processing. They are capable of generating realistic images and have shown great potential in tasks such as image synthesis, editing, and enhancement. Continuous innovation and optimization of neural network structures, such as such as GAN and Transformer, have led to remarkable achievements in image processing tasks. This report introduces the applications of image generation technologies in image colorization and fusion. Due to the inherent learning mechanism of deep learning models, they may be misled or even manipulated. Thus, image generative models face security issues, and themselves may generate security generation threats as well, particularly, the issues such as adversarial attacks against models and forged image. This report discuses the adversarial attack for cross-task multi-model system based on the research of adversarial attack and defense for single-task model, and the detection technology for generated images is also discussed.
Biography:
Xin Jin received his BS degree in electronics and information engineering from the Henan Normal University, Xinxiang, China, in 2013, and his PhD in Communication and Information Systems from Yunnan University, Kunming, China, in 2018. He was a Post-Doctoral Fellow with the School of Software, Yunnan University from 2018 to 2020. He has published over 100 articles in IEEE Transactions on Cybernetics, IEEE Transactions on Instrumentation & Measurement, IEEE Transactions on Geoscience and Remote Sensing, IEEE Transactions on Consumer Electronics, IEEE Transactions on Emerging Topics in Computational Intelligence, IEEE Transactions on Intelligent Vehicles , Engineering Applications of Artificial Intelligence, Information Sciences, Expert Systems with Application and so on. He has more than 1400 citations, and his H-Index is 19. He has won the prize of Natural Science Award of Yunnan Province Science and Technology Award, the prize of Excellent Achievement Award of Yunnan Province Postdoctoral Research Fund, and the Excellent Doctoral Thesis of Yunnan Province. He served as the Guest editor of Frontiers in Neurorobotics, Life, Discover Applied Sciences, and Intelligence & Robotics.