As U.S. DOE has moved toward data-driven scientific discovery, machine learning (ML) has become a critical technology in the modeling of complex phenomena in concert with current computational, experimental, and observational approaches. However, development of ML approaches for many scientific domains poses several challenges such as data paucity, domain-knowledge integration, and adaptability. In this talk, we will discuss our work on scientific domain-informed ML approaches that seek to overcome these challenges. We will present our recent work on image-based deep-learning methods for compressed sensing and segmentation, diffusion-process-based deep-learning method for large-scale transportation network modeling, and reinforcement learning approaches for automated scientific ML. We will conclude with some exciting avenues for future research in scientific ML.

Argonne Physics Division Colloquium Schedule