Computer Vision News - June 2023

38 ICLR 2023 Workshop uncertainty evaluation in the talk “ Safely Utilizing AI Model in Open Clinical Environment ”. Shandong Wu discussed the trustworthiness of AI technology from a more clinical perspective in the talk “ Trustworthy Medical AI in the Loop of AlgorithmandClinic ”,whereheemphasized that the AI algorithms and clinic needs mutually fostering each other. Georgios Kaissis shared their works on protecting data privacy by “ Unlocking the Potential of Differential Privacy in Medical Imaging: Enabling Data Analysis while Protecting Patient Privacy ”. Meanwhile, Xiaoxiao Li , who has also been working on privacy- preserving machine learning, discussed their effort on federated learning with the talk “ Overcoming Data Heterogeneity current medical imaging procedures and presented possible solutions. “ Standardized and transparent evaluation is one of the keys to successfully deploying any healthcare solutions into practice, to make sure that an AI product is robust, fair, and trustworthy for clinical use. ” Luyang told us. The workshop also has five other invited talks. Minghao Cheng shared their effort on visualizing a global explanation in the input space for every class learned in the training procedure by his talk “Generating Class-wise Visual Explanations for Deep Neural Networks”. Huazhu Fu introduced how to develop reliable and robust machine learning algorithms with Minghao Cheng Huazhu Fu Shandong Wu Georgios Kaissis Xiaoxiao Li