4:30–6:00 pm ERC 401
Aleksandra Ćiprijanović (Fermilab) "Bridging the Gap Between Astronomical Datasets: From Proof of Concept to AI Model Deployment with Domain Adaptation"
Artificial intelligence methods show great promise in increasing the quality and speed of work with large astronomical and high-energy physics datasets. The high complexity of these methods leads to the extraction of dataset-specific, non-robust features, hence models do not generalize well across multiple datasets. As proof of concept, deep learning models are often trained on simulations with the prospect of being deployed and used on real data in the future. Unfortunately, this often leads to a substantial decrease in model performance on the real data. In this talk I will introduce “the domain shift problem” and why it appears in the sciences. Finally, I will introduce methods to overcome this problem and show several example studies performed by our group. With further development, these techniques will allow scientists to construct deep learning models that can successfully combine the knowledge from simulations and real data originating from multiple instruments.
Aleksandra Ćiprijanović, PhD:Aleksandra is a Wilson Fellow Associate Scientist at the Data Science, Simulation, and Learning Division at Fermilab and is also leading the Cosmic AI group. Before this position, she was an Assistant Research Professor at the University of Belgrade, Serbia, and the Mathematical Institute, Serbian Academy of Sciences and Arts. She is interested in the formation and evolution of structures in the Universe – from galaxies and galaxy clusters to large-scale structures. Her work focuses on advancing and building trustworthy and robust AI algorithms that will allow us to fully utilize all available data in the era of large astronomical surveys.