With increasingly better instruments, we can now see things at atomic scales, measure vibrations imperceptible to the human eye, and capture high-resolution images of objects millions of light-years away. But those instruments are also producing vastly more data than ever before. Machine learning methods tailored to scientific data provide powerful tools for analyzing these complex datasets, as well as controlling scientific experiments. 

Berkeley Lab researchers have been addressing these challenges for the last several years, and from August to September 2020, we featured some of their projects and postdocs in a social media campaign. Here’s a collection of that content. For more on machine learning for science research at Berkeley Lab visit: https://ml4sci.lbl.gov/.