Computer Vision News - June 2023

39 Trustworthy ML for Healthcare strongly motivated this workshop to share, discuss, and explore trustworthy machine learning algorithms for a more practical clinic use in the future.” Also, aspart of thepost-workshopactivities, Hao and Luyang are co-organizing a special issue on the IEEE Journal of Biomedical and Health Informatics (J-BHI) entitled “ TrustworthyMachine Learning for Health Informatics ” . “The 1st TML4H workshop is just a beginning, and we look forward to more fruitful studies on the special issue as well as the next workshop!” Challenges in Federated Learning ”. “The discussion in this workshop was versatile.” Luyang said, “Apart from the keynote and the invited talks, the oral presentations covered various aspects of trustworthiness, such as explainability, privacy, cross-domain robustness , etc. Many papers are related to explainability, and the best paper selected by the program committee is also on this topic. Other papers also impressed us a lot, from which we selected two best paper honorable mentions.” “What distinguishes our workshop from others is that we are trying to draw the attention of the research community on the important topic of trustworthy ML. We are indeed facing such a challenge. For example, a paper published at the Nature Machine Intelligence on 2021 reviewed 62 studies developing AI models for COVID-19 analysis and found that none of them are of clinical usage due to methodological flaws and/or underlying biases 1 . This 1. Roberts, Michael, et al. "Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans." Nature Machine Intelligence 3.3 (2021): 199-217.