Archimedes' Summer Activities kicked off with our successful Workshop on AI in Imaging!
This past Tuesday, May 28th 2024, Archimedes successfully kicked off our summer activities with an engaging workshop held at our local Amphitheater in Marousi, Athens. The event, which offered both in-person and hybrid attendance, featured a series of insightful presentations by renowned experts in artificial intelligence, computer vision, and precision medicine. The workshop attracted a diverse audience eager to explore the latest advancements and research in these dynamic fields.
The day began at 9:30 AM with a warm welcome and coffee, providing attendees an opportunity to network and set the tone for the day's sessions. At 10:00 AM, the first presentation, "Conditioning Generative Models," was delivered by Dimitris Samaras. Samaras, a SUNY Empire Innovation Professor at Stony Brook University, delved into conditioning generative models without retraining them by using posterior inference. He showcased the application of self-supervised learning features in digital histopathology, highlighting their effectiveness in the absence of detailed text annotations, and presented compelling examples of controlling 3D generative models.
Following a brief question and break period, Vicky Kalogeiton, an Assistant Professor at École Polytechnique, took the stage at 10:50 AM with her talk on "Multimodality in Computer Vision." Kalogeiton emphasized the significance of using multiple modalities—such as video, audio, text, and clinical data—for visual recognition and generation. She illustrated her points with examples from video understanding and real clinical data applications, demonstrating the transformative potential of multimodal approaches in medical imaging and beyond.
After another short break, Maria Vakalopoulou, an Assistant Professor at CentraleSupelec University Paris-Saclay, presented her research on "Precision Medicine, AI, and Longitudinal Monitoring" at 11:40 AM. Vakalopoulou explored the integration of longitudinal data in medical settings, particularly in radiology for lung and kidney transplants. Her discussion on foundation models and novel formulations for handling imaging and multimodal data provided valuable insights into the future of AI in medical diagnostics and treatment planning.
The final presentation of the workshop, "Parameter Efficient Fine Tuning of Large Foundation Models in Precision Medicine," was given by Stergios Christodoulidis, Assistant Professor at CentraleSupélec, University Paris-Saclay, at 12:30 PM. Christodoulidis addressed the challenges of adapting large models to new domains with limited training data. He highlighted probabilistic formulations for creating reliable, well-calibrated models capable of providing uncertainty quantifications, a crucial aspect for advancements in personalized medicine.
Throughout the workshop, each presentation was followed by a brief question and break period, fostering deeper interaction between the speakers and the audience. Archimedes is looking forward to continuing this momentum with future events, aimed at furthering knowledge exchange and fostering innovation. The event concluded at 1:10 PM, marking an exciting start to Archimedes' summer schedule. The high-quality presentations and the engaging hybrid format set a positive tone for upcoming events. Archimedes continues to be a pivotal platform for collaborations that drive advancements in technology and AI applications across various fields.
