3:30–4:30 pm MCP 201
Machine learning has the potential to transform scientific research. This fundamental change cannot be realized through the straightforward application of existing off-the-shelf machine learning tools alone. Rather, we need novel methods for incorporating physical models and constraints into learning systems, generating hypotheses of physical laws governing observed data, designing mechanisms for data collection, leveraging and accelerating simulations, and quantifying the uncertainty of models’ predictions. In this talk, I will describe opportunities and emerging tools for addressing these challenges. We will see how ideas from statistics, optimization, scientific computing, and signal processing are informing modern machine learning methods used across the natural sciences.