Image Synthesis
Image synthesis focuses on generating realistic images from various inputs, such as text descriptions, sketches, or other images, aiming to improve controllability, realism, and efficiency. Current research emphasizes advancements in diffusion models, generative adversarial networks (GANs), and autoregressive models, often incorporating techniques like latent space manipulation, multimodal conditioning (text and image), and attention mechanisms to enhance image quality and control. This field is significant for its applications in diverse areas, including medical imaging, virtual try-ons, and content creation, while also raising important considerations regarding ethical implications and environmental impact of computationally intensive models.
Papers
DPAF: Image Synthesis via Differentially Private Aggregation in Forward Phase
Chih-Hsun Lin, Chia-Yi Hsu, Chia-Mu Yu, Yang Cao, Chun-Ying Huang
FIANCEE: Faster Inference of Adversarial Networks via Conditional Early Exits
Polina Karpikova, Radionova Ekaterina, Anastasia Yaschenko, Andrei Spiridonov, Leonid Kostyushko, Riccardo Fabbricatore, Aleksei Ivakhnenko