Modern spectroscopy techniques can produce tens of thousands of spectra from a single experiment, which has placed a considerable burden on traditional human-driven methods for interpretation of these spectra. A research team led by The University of Tokyo Institute of Industrial Science combined two machine learning techniques, layer clustering and decision tree methods, to produce data-driven methods for spectral interpretation and prediction that can analyze any spectral data quickly and accurately.

The largest continuous 3D color reconstruction of a seafloor area was just generated at Hydrate Ridge South Summit aboard Schmidt Ocean Institute’s research vessel Falkor using data processing algorithms.